268 chapter 3 surgical complications and consequently additional costs. Therefore, an additional diagnostic test or combination of tests may prevent unbeneficial diagnostic hemithyroidectomies for benign nodules by ruling out malignancy, and/or prevent two-stage surgery if malignancy can be confirmed pre-operatively. A multimodal stepwise approach using a sensitive rule-out test and a specific rule-in test might provide the most conclusive diagnosis for indeterminate thyroid nodules [25]. The ATA guidelines state that an ideal ‘‘rule-in’’ test would have a positive predictive value (PPV) similar to a malignant cytologic diagnosis (i.e., 98.6%), and an ideal ‘‘rule-out’’ test would have a negative predictive value (NPV) similar to a benign cytologic diagnosis (i.e., 96.3%) [17, 49]. Diagnostic tests on cytological samples might include molecular tests like gene mutation panels, gene or microRNA expression profiles, immunocytochemistry, and sequencing techniques. Also, several imaging modalities may be used in the workup of indeterminate thyroid nodules in vivo, including anatomical imaging techniques such as US and magnetic resonance imaging (MRI), and molecular imaging techniques such as 2-[18F]fluoro-2-deoxy-D-glucose ([18F]FDG) positron emission tomography (PET) combined with computed tomography (CT) and hexakis(2-methoxy-2-methylpropylisonitrile)technetium[99mTc] ([99mTc]Tc-MIBI, also known as [99mTc]Tc-sestaMIBI) scintigraphy with single photon emission computed tomography combined with CT (SPECT/CT). Performance of any diagnostic tests are usually expressed by their sensitivity, specificity, PPV, NPV, accuracy, diagnostic odds ratio (dOR), positive likelihood ratio (LR+), negative likelihood ratio (LR-), area under the receiver operating characteristic curve (AUC), and benign call rate, when available. The reader should be aware that, albeit clinically very useful, parameters like PPV, NPV but also accuracy and benign call rate are dependent on the a priori risk of a patient to suffer from the disease and thus can only be compared between different cohorts if the definition and prevalence of the malignancy are similar. Most studies featured in this chapter present outcome measures of a single cohort, without validation in a separate part of the dataset or (preferably) in an external cohort. When (external) validation was performed, this is specifically mentioned. This chapter provides an overview of biomarkers obtained using conventional as well as AI-based non-invasive imaging strategies for the differentiation of thyroid nodules with indeterminate cytology. It presents the ability of a test to differentiate between benign and malignant nodules, taking into account the clinical readiness and cost-effectiveness. This chapter presents studies with different definitions of indeterminate cytology: some include Bethesda III and IV, only; others also include Bethesda V; and some have not incorporated the Bethesda System yet. The definition of indeterminate cytology will be specifically reported, when available.
RkJQdWJsaXNoZXIy MTk4NDMw