Tracheal invasion by thyroid carcinoma: prediction using MR imaging.

OBJECTIVE: The purpose of this study was to evaluate the accuracy of MR imaging in predicting tracheal invasion by thyroid carcinomas and to determine MR imaging criteria for diagnosing tracheal invasion.
MATERIALS AND METHODS: MR imaging was performed on the normal trachea of one cadaver and 30 healthy subjects as a standard of reference. Then, MR imaging findings in 67 patients with thyroid carcinoma were reviewed and correlated with surgical and pathologic findings. A logistic regression model was used to determine which MR imaging features were significant for predicting tracheal invasion.
RESULTS: Twenty-three (34%) of the 67 patients had tracheal invasion. Logistic regression model analysis revealed that significant MR characteristics for determining tracheal invasion included soft-tissue signal in the tracheal cartilage (p < 0.001), intraluminal mass (p < 0.001), and degree of tumor circumference around the trachea (p = 0.001). The highest accuracy (90%) for determining tracheal invasion was achieved using a combination of findings. A case was considered positive for tracheal invasion if there was soft-tissue signal in the cartilage, an intraluminal mass, or a tumor that abutted a circumference of the trachea of 180 degrees or greater. Using these factors resulted in seven false-positive diagnoses because soft-tissue signal in the cartilage was sometimes seen in healthy trachea. Although intraluminal mass invariably reflected deep tracheal invasion, soft-tissue signal in the cartilage rarely indicated actual cartilage invasion but rather indicated tumor extension between the cartilaginous rings.
CONCLUSION: Tracheal invasion by thyroid carcinomas can be accurately diagnosed with MR imaging, and using a combination of criteria is the most accurate method of predicting this phenomenon.