A MicroRNA based thyroid molecular classifier for thyroid nodules: a real-world independent study

Abstract

Thyroid nodules are common and affect around 50% of individuals. These nodules are often discovered incidentally and exhibit benign characteristics. Following a suspicious ultrasound, a fine needle aspiration biopsy (FNAB) assesses the risk of malignancy. However, approximately 30% of cases are classified as indeterminate by cytology. In response, the development of molecular tests has helped refine malignancy risk and reduce diagnostic surgeries. The present study aimed to evaluate the performance of a microRNA-based molecular test (mir-THYpe®) in improving diagnostic accuracy in indeterminate thyroid nodules. This retrospective, observational, and non-interventional study included patients who underwent the genetic test in the state of Santa Catarina, Brazil. A total of 256 patients with nodules classified as Bethesda III/IV were analyzed. The test was positive for malignancy in 90 patients, 79 (90%) of whom underwent surgery. Of the 158 test-negative patients, 7 (4.4%) underwent thyroidectomy. Since not all test-negative nodules could be assumed to be truly benign, the sensitivity from the validation study, based on Bayes' theorem, was applied. The test demonstrated a sensitivity of 83.0%, a specificity of 83.5%; a positive predictive value (PPV) of 62.8% and a negative predictive value (NPV) of 93.6%. In this study, the mir-THYpe test supported 95.5% of clinical decisions when negative and 90.1% when positive, leading to a surgery avoidance rate of 61.2%. Therefore, the integration of a microRNA-based classifier into clinical practice is a valuable tool in managing indeterminate thyroid nodules and reducing unnecessary thyroidectomies.