Can a CT scan distinguish recurrent thyroid cancer from thyroid tissue left after surgery?

Details

Thyroidectomy can leave some thyroid tissue behind. This complicates imaging surveillance because residual tissue must be distinguished from local recurrence. The study investigated whether multiphasic multi-detector computed tomography (4D-MDCT) could distinguish residual nonmalignant thyroid tissue from recurrent thyroid carcinoma after thyroidectomy.

In this retrospective study, researchers compared Hounsfield unit values on multiphasic multidetector CT in the precontrast, arterial (25 seconds), venous (55 seconds), and delayed (85 seconds) phases across 29 recurrent thyroid cancer lesions, 29 normal thyroid samples, and 29 diseased thyroid samples with thyroiditis or multinodular thyroid. They used ANOVA to compare Hounsfield unit values among lesion types in each phase. Logistic regression and receiver operating characteristic analysis were used to evaluate how well Hounsfield unit values predicted recurrence.

All three tissue types showed nearly parallel enhancement characteristics, with a wash-in–washout pattern. Hounsfield unit density differed statistically across all four phases: recurrence had the lowest values, diseased thyroid intermediate values, and normal thyroid the highest values (P < .001 for each group and in each phase). In a univariate logistic regression analysis comparing recurrence against diseased or normal thyroid tissue, the precontrast phase had the greatest area under the receiver operating characteristic curve for distinguishing benign from malignant thyroid tissue: 0.983 (95% CI, 0.954–1), at a cutoff of ≤62 Hounsfield units, with sensitivity of 0.966 and specificity of 0.983; the arterial phase followed.

The authors concluded that multiphasic multidetector CT can distinguish recurrent thyroid carcinoma from residual nonmalignant thyroid tissue with high accuracy. The precontrast images provided the most information for discrimination, followed by the arterial phase. They suggested that an optimal clinical protocol could use any number of phases but should include a precontrast phase.