Predictive Value of Serum Calcitonin Levels for Preoperative Diagnosis of Medullary Thyroid Carcinoma in a Cohort of 5817 Consecutive Patients with Thyroid Nodules

Abstract
Routine serum calcitonin (CT) measurement in patients with thyroid nodules for diagnosis of medullary thyroid carcinoma (MTC) is controversial. The objective of this study was to evaluate the diagnostic accuracy of systematic CT measurement in non-multiple endocrine neoplasia type 2 patients with nodular thyroid disease. This study was conducted at a national healthcare system hospital (outpatient and inpatient sectors). Consecutive patients with nodular thyroid disease (n = 5817) were studied. Serum CT levels were measured under basal conditions, and when basal values were more than or equal to 20 and less than 100 pg/ml, testing was repeated after pentagastrin stimulation. Basal or stimulated levels more than 100 pg/ml were indication for surgery. Fifteen cases of MTC and seven of C cell hyperplasia (CCH) were identified. MTCs were diagnosed in all patients with basal CT more than 100 pg/ml. The four patients with basal CT more than or equal to 50 and less than 100 pg/ml included two diagnosed with MTC and two with CCH. In 10 patients with basal levels more than or equal to 20 and less than 50 pg/ml, histology confirmed the presence of MTC in four, four others had CCH, and the remaining two were negative for thyroid malignancy. Positive predictive values for basal CT levels in the preoperative diagnosis of MTC were: 23.1% for values more than or equal to 20 pg/ml, 100% for values more than 100 pg/ml, 25% for levels more than or equal to 50 and less than 100 pg/ml, and 8.3% for values more than or equal to 20 and less than 50 pg/ml. Positive predictive values for the pentagastrin test (>100 pg/ml) were 40% in the entire series. CT screening of thyroid nodules is a highly sensitive test for early diagnosis of MTC, but confirmatory stimulation testing is necessary in most cases to identify true positive increases.

This publication has 24 references indexed in Scilit: