Hypoxia-associated gene signature for uterine cervical cancer

Abstract Background Tumour hypoxia is a major determinant of treatment resistance and poor prognosis in cervical cancer but remains difficult to assess in clinical practice. Gene expression signatures offer a potential means to characterise hypoxia-related biology. This study aimed to develop and validate a hypoxia-associated gene expression signature for cervical cancer. Methods RNA sequencing was performed on five cervical cancer cell lines exposed to normoxia (21% O?) and hypoxia (1% O?). Differentially expressed genes were mapped to The Cancer Genome Atlas cervical cancer cohort (TCGA-CESC) to train a 55-gene hypoxia classifier using k-means clustering and Prediction Analysis for Microarrays. The model was validated in an institutional Manchester cohort (n=153) and two public datasets from Seoul (n=300) and Oslo (n=283). Results The Manchester 55-gene signature was enriched for canonical hypoxia pathways. In the Manchester cohort, hypoxia classification correlated with advanced FIGO stage, nodal involvement, tumour size ? 4 cm, and hydronephrosis (adjusted p<0.05). Hypoxic tumours showed reduced overall survival (OS) and progression-free survival (PFS) in all cohorts. In multivariable models, the signature remained independently prognostic for OS in both TCGA (HR 1.70, 95% CI 1.10-2.60, p=0.012) and Manchester (HR 1.95, 95% CI 1.08-3.51, p=0.026). A direct comparison with a published 6-gene hypoxia signature in the Oslo cohort demonstrated 71% concordance in classification. Conclusions Our 55-gene signature should be tested prospectively in trials to assess its ability to stratify patients for hypoxia-targeted therapies.