Abstract
Endogenous human retroviruses (ERVs) are remnants of exogenous retroviruses that have integrated into the human genome. Using publicly available RNA-seq data from 63 cervical cancer patients, we investigated the expression of ERVs in cervical cancers. Four aspects of cervical cancer were investigated: patient ancestral background, tumor HPV type, tumor stage and patient survival. Between the racial subgroups, 74 ERVs were significantly differentially expressed, with Black Americans having 30 upregulated and 44 downregulated (including MER21C, HERV9-int, and HERVH-int) ERVs when compared to White Americans. We found that 3313 ERVs were differentially expressed between HPV subgroups, including MER41A, HERVH-int and HERVK9. There were 28 downregulated (including MLT1D and HERVH-int) and 61 upregulated (including MER41A) ERVs in locally advanced-stage compared to early-stage samples. Tissue microarrays of cervical cancer patients were used to investigate the protein expression of ERVs with protein coding potential (i.e., HERVK and ERV3). Significant differences in protein expression of ERV3 (p = 0.000905) were observed between early-stage and locally advanced-stage tumors. No significant differential expression at the protein level was found for HERVK7 (p = 0.243). We also investigated a prognostic model, supplementing a baseline prediction model using FIGO stage, age and HPV positivity with ERVs data. The expression levels of all ERVs in the HERVd were input into a Lasso-Cox proportional hazards model, developing a predictive 67-ERV panel. When ERVs expression levels were supplemented with the clinical data, a significant increase in prognostic power (p = 9.433 × 10(-15)) relative to that obtained with the clinical parameters alone (p = 0.06027) was observed. In summary, ERV RNA expression in cervical cancer tumors is significantly different among racial cohorts, HPV subgroups and disease stages. The combination of the expression of certain ERVs in cervical cancers with clinical factors significantly improved prognostication compared to clinical factors alone; therefore, ERVs may serve as future prognostic biomarkers and therapeutic targets. Novelty and Impact: When endogenous retroviral (ERV) expression signatures were combined with currently employed clinical prognosticators of relapse of cervical cancer, the combination outperformed prediction models based on clinical prognosticators alone. ERV expression signatures in tumor biopsies may therefore be useful to help identify patients at greater risk of recurrence. The novel ERV expression signatures or adjacent genes possibly impacted by ERV expression described here may also be targets for the development of future therapeutic interventions.