Development and validation a nomogram based on pathological microscopic features to predict survival in nasopharyngeal carcinoma and guide treatment decision

Abstract Background In various types of solid tumor, pathological microscopic features can be used as prognostic biomarkers. We intended to develop a pathological microscopic features (PMFs) pattern for individual survival assessment in patients with nasopharyngeal carcinoma (NPC). Methods We retrospectively included 1229 patients with NPC who had received radical radiotherapy with/without chemotherapy. The PMFs differently digitized were extracted using the software QuPath (version 0.1.2.Queen’s University Belfast, Belfast, Northern Ireland, UK) in the training cohort (Guangzhou training cohort, n = 739) to bulid a pathological feature classifier using a penalized regression model. The prognostic accuracy of the pathological feature classifier was validated in the internal validation cohort (Guangzhou validation cohort, n = 316) and one external validation cohort (Chenzhou validation cohort, n=170).The primary end point was progression-free survival (PFS) and distant metastasis-free survival (DMFS). Results We found 143 PMFs in the HE p = 0.000), and DMFS (HR 1.87, 1.27 -2.75; p = 0.001) than patients with low-risk scores. We developed a nomogram based on the PMSRN and other variables that predicted an individual’s survival risk and the c-index of this nomogram (C-index=0.759, 95% CI: 0.717 -0.801) was equal to other nomograms. Furthermore, among patients with high-risk scores in the combined training and internal cohorts, ICT had better PFS to those who received CCRT alone (p = 0.033), whereas those with low-risk scores ICT had similar PFS to CCRT alone (p = 0.363). These results were validated in the internal external validation cohort. Conclusions The PMSRN is a reliable prognostic tool for survival risk in NPC patients and might be able to predict which patients need to receive ICT or not. It might guide treatment decisions for NPC patients. Clinical trial identification ChiECRCT20190034. Legal entity responsible for the study Sun Yet-sen University Cancer Center. Funding National Natural Science Foundation of China. Disclosure All authors have declared no conflicts of interest.