There are four most commonly used clinical prediction models that estimate the probability of malignancy in pulmonary nodules. While three models are based on clinical and computed tomography (CT) characteristics (Mayo Clinic, Veterans Association, Br...
There are four most commonly used clinical prediction models that estimate the probability of malignancy in pulmonary nodules. While three models are based on clinical and computed tomography (CT) characteristics (Mayo Clinic, Veterans Association, Brock University), one model additionally used positron emission tomography (PET) characteristics (Herder model). This study aimed to compare these four models at our institution. Retrospective analysis of 242 patients with pathologically confirmed nodules (4-30 mm in diameter) from January 2015 to December 2015 was performed. Receiver operating characteristic area under the curve (AUC) analysis was used to assess the predicting accuracy for malignancy. Of 242 patients, 187 patients (77.2%) had malignant nodules and 55 patients (22.8%) had benign nodules. PET was performed in 236 patients (97.5%). The Mayo, Brock and Veterans Association (VA) models showed similar performance (AUC 0.6337 vs. AUC 0.7138 vs. AUC 0.6121 respectively). Accuracy of the Herder model (AUC 0.6429) was not significantly different from those of the other three models (Mayo vs. Herder P=0.8262, Brock vs. Herder P=0.1435, VA vs. Herder P=0.5581). There were no differences among four models in determining probability of malignancy in pulmonary nodules. The usefulness of PET in predicting malignancy was limited in our study.