Development And Validation Of Prognostic Nomogram Based On Tumor Residual After Radiotherapy In Patients With Nasopharyngeal Carcinoma

Background:Patients with nasopharyngeal carcinoma（NPC）often have tumor residual after radiotherapy（RT）.There are still many controversies about the approach and timing of diagnosing residual tumor.A number of studies have shown patients with residual tumor had significantly worse prognosis than those of without residual tumor.There is currently no uniform standard for residual tumor treatment.The nomogram model has been used in a variety of tumor to facilitate further prognostic stratification and to guide clinical decision-making.Objective:The purpose of this study was to further investigate the prognosis value of patients with nasopharyngeal carcinoma who were diagnosed with tumor residual and non-tumor residual by magnetic resonance imaging（MRI）.A nomogram model for predicting prognosis is constructed based on tumor residual to help guide clinical decision-making.Methods:The study retrospectively collected clinical and follow-up data from patients with nasopharyngeal carcinoma who underwent radiotherapy at our hospital from January 2008 to October 2017.A total of 538 patients without metastatic,histologically diagnosed nasopharyngeal carcinoma were used for retrospective analysis.All patients underwent MRI examination of the nasopharynx and neck before and after enrollment.According to the magnetic resonance results after radiotherapy completion 6 months,the patients were divided into residual tumor cohort and non-residual tumor cohort.Kaplan-Meier curve was used to compare the difference in prognosis between residual tumor and the non-residual tumor cohort.Of the above patients,379 patients were used to construct and validate the nomogram model including training set（230 participants）and the internal validation set（149 participants）.Combined with the multivariate analysis results of the training set and previous literature reports,a predictive prognostic nomogram model was created using the rms installation package of the R software,and verified in the internal validation group.C-index（concordance index）was used to evaluate the accuracy and discrimination ability of the nomogram model.Result:In general,the 3-year progress-free survival（PFS）rate was 69.4%,3-year locoregional recurrence-free survival（LRFS）rate was 78.7%,3-year distant metastasis-free survival（DMFS）rate was 78.4%.The 3-year rates for the non-residual tumor and residual tumor cohort were:PFS（73.4%VS 61%,P=0.009）.LRFS（81.9%VS 72%,P=0.02）,and DMFS（80.7%VS 73.5%,P=0.11）.9 significant factors were identified and included into the nomogram model.They were N stage,T stage,age（<60,≥60）,RT technique,histologic classification,sex,induction chemotherapy,maximum diameter of local residual tumor,EB V DNA before treatment respectively.The calibration curve for the probability of progress-free survival showed that the nomogram-based predicting was in good consistent with actual observation in the internal validation set,the C-index was 0.71（95%CI 0.418-0.772）.Conclusion:MRI-detected residual tumor after RT was associated with prognostic negatively in NPC.The proposed nomogram served as a potential tool to predict prognosis,guiding therapeutic strategies and aiding clinicians decision-making concerning local residual.Some effective and timely strengthening treatment need to be administered to patients with high possibility to progress.