Validation and Comparison of Nomograms in Predicting Disease-Specific Survival for Papillary Thyroid Carcinoma.

Abstract

Nomogram could estimate individualized prognosis in papillary thyroid carcinoma (PTC). We aimed to create and validate a new nomogram and compare it with other published nomograms using a large patient cohort. Eight-hundred and forty-nine PTC patients with ≥7 years follow-up were randomly assigned to the development (n=425) and validation (n=424) groups. The former was used for developing a nomogram for disease-specific survival (DSS), while the latter was for validating the nomogram by discrimination [or area under curve (AUC)]. AUC of the newly developed nomogram was compared to other published nomograms. The 5- and 10-year risk of dying from PTC were 1.4 and 3.3%, respectively, while dying from non-PTC-related causes were 2.3 and 5.1%, respectively. The new nomogram was developed from age, tumor size, multifocality, nodal status and distant metastases. The discrimination was excellent (AUC (95% CI) for 5- and 10-year DSS were 0.896 (0.683-0.971) and 0.919 (0.871-0.967), respectively). Its predictability was similar to other published nomograms (p>0.05). Based on the new nomogram, a total score of <28 meant 99.72% chance of surviving from PTC at 10 years while a score of ≥28 meant 9.09% chance of dying from PTC at 10 years. Using variables from the current tumor node metastasis (TNM) staging system, a new nomogram was developed. It exhibited excellent discriminatory ability and accuracy in predicting 10-year DSS relative to other published nomograms. However, given the excellent prognosis of PTC, the new nomogram was better at ruling out than predicting PTC-related death. Further validation by an external cohort is required.