摘要
Objectives: To develop and validate a prognostic model of integrating whole-tumour apparent diffusion coefficient (ADC) from pretreatment diffusion-weighted (DW) magnetic resonance (MR) imaging with human papillomavirus (HPV) genotyping in predicting the overall survival (OS) and disease-free survival (DFS) for women with stage IB–IV cervical cancer following concurrent chemoradiotherapy (CCRT). Methods: We retrospectively analysed three prospectively collected cohorts comprising 300 patients with stage IB–IV cervical cancer treated with CCRT in 2007–2014 and filtered 134 female patients who underwent MR imaging at 3.0 T for final analysis (age, 24–92 years; median, 54 years). Univariate and multivariate Cox regression analyses were used to evaluate the whole-tumour ADC histogram parameters, HPV genotyping and relevant clinical variables in predicting OS and DFS. The dataset was randomly split into training (n = 88) and testing (n = 46) datasets for construction and independent bootstrap validation of the models. Results: The median follow-up time for surviving patients was 69 months (range, 9–126 months). Non-squamous cell type, ADC <sub>10</sub> <0.77 × 10 <sup>-3</sup> mm <sup>2</sup> /s, T3-4, M1 stage and high-risk HPV status were selected to generate a model, in which the OS and DFS for the low, intermediate and high-risk groups were significantly stratified (p < 0.0001). The prognostic model improved the prediction significantly compared with the International Federation of Gynaecology and Obstetrics (FIGO) stage for both the training and independent testing datasets (p < 0.0001). Conclusions: The prognostic model based on integrated clinical and imaging data could be a useful clinical biomarker to predict OS and DFS in patients with stage IB–IV cervical cancer treated with CCRT. Key points: • ADC <sub>10</sub> is the best prognostic factor among ADC parameters in cervical cancer treated with CCRT • A novel prognostic model was built based on histology, ADC <sub>10</sub> , T and M stage and HPV status • The prognostic model outperforms FIGO stage in the survival prediction.