A comparison of prognostic classification systems for differentiated thyroid carcinoma

Abstract

Objective To identify and compare prognostic classification systems based on basic tumour characteristics that were developed and/or validated for differentiated thyroid carcinoma (DTC). Design Retrospective chart study. Methods Literature was studied using PubMed. Fifteen different prognostic classification systems were identified, of which seven were developed or validated for DTC patients and were based on basic tumour characteristics. These systems were applied to 1225 DTC patients who were treated in our hospital between 1978 and 2002. Results Log-rank analysis of Kaplan-Meier cancer-specific survival curves showed that the curve of the Tumor, Lymph-Node, Metastasis (TNM) system had the greatest discriminatory power (log-rank test, log P = -84·9). Cox-regression analysis showed that the TNM system was the most powerful determinant of cancer-specific survival curves. Proportion of variance explained (PVE) analysis showed that the TNM system had the highest PVE. Conclusion Of the prognostic classification systems analysed in this study, the TNM system performs best in all the three analyses, and is therefore the most suitable for predicting outcome in DTC-patients. © 2010 Blackwell Publishing Ltd.