ATF4, ZIK1, and PNMA3 Transcriptomic Signatures Predict Chemoradiotherapy Response in Cervical Cancer.

BACKGROUND: Cervical cancer (CC) is a main malignancy affecting women globally, with a high mortality rate. Chemoradiotherapy resistance is a significant factor in determining treatment outcomes, and prognostic biomarkers are essential for the appropriate selection of therapeutic strategies. This study aimed to identify gene expression signatures that may allow for follow-up treatment response in CC patients. METHODS: Samples of non-stem CC cells selected by Fluorescence-activated Cell Sorting from 21 Responders and 10 Non-responder patients, who attended the Mario Penna Institute (Belo Horizonte-Brazil) from August 2017 to May 2018, and submitted to Next-generation RNA sequencing, were analyzed to evaluate differential gene expression profiles between both groups. Machine Learning and statistical analyses were performed to identify a signature related to clinical outcomes. RESULTS: 2670 Differentially expressed genes (DEGs) with log2FoldChange ≤ -1 or ≥ 1 and adjusted p-value ≤ 0.05 were found. Of these, 1591 were protein-coding, but only 8 were considered relevant based on machine-learning and statistical analyses for classifying treatment response. However, only PNMA3 and ZIK1 were markedly upregulated in non-responder patients. Among the genes upregulated in Responder patients, only ATF4 and ATP1B3 were identified by machine-learning analysis. On the other hand, only ATF4, PNMA3, and ZIK1 were associated in network interactions, more specifically to apoptosis pathways, and the differential expression of ATF4 and PNMA3 occurred exclusively in cervical cancer non-stem cells, in the presence of ZIK1. CONCLUSIONS: We propose that upregulation of PNMA3 and downregulation of ATF4, specifically in cervical cancer non-stem cells, is associated with upregulation of ZIK1 and with resistance to apoptosis and chemoradiotherapy. Therefore, suggesting that these genes may serve as potential signatures for CC prognosis.