71 Diagnostic utility of molecular and imaging biomarkers 2 features justifies the special attention given to Hürthle cell cytology by the Bethesda and other classification systems. An accurate additional diagnostic is desired. Disappointingly, several studies concluded that the investigated test was accurate in all except Hürthle cell lesions [108, 170, 296, 309]. Immunocytochemistry handed some solutions, although promising results of combined galectin-3 and CK-19 staining have not yet been validated [212]. Besides that, BRAF, RAS, RET/PTC or PAX8/PPARγ alterations are only occasionally found [108, 114]. These findings support previous presumptions that oncocytic thyroid nodules are a completely separate entity with a unique molecular and phenotypic profile [339-342]. Malignant transition in Hürthle cell nodules most likely involves the PIK3CA-Akt-mTOR and Wnt/bèta-catenin pathways rather than the MAPK/ERK pathway [339, 342]. Rare TP53 mutations, usually associated with poorly differentiated and anaplastic carcinoma, were recently also identified in well-differentiated Hürthle cell nodules [340]. Also, recurrent FTC-OV have shown genome haploidization, a rare phenomenon in other types of differentiated thyroid carcinoma [343]. Specific markers for the preoperative molecular differentiation of Hürthle cell nodules should be developed. Adaptation of existing tests to additionally suit Hürthle cell nodules (e.g. the Afirma® GEC) is a strategy being explored, for example by the ThyroSeq® v3 and the Afirma® Gene Sequence Classifier. Caution should be taken that these adaptations do not decrease the diagnostic accuracy for non-oncocytic lesions. MicroRNA expression profiling of these lesions is currently also under investigation [186, 190]. Strengths and limitations of the current review There are several important strengths and limitations to this comprehensive review. This review provides a complete overview of the available additional diagnostics for indeterminate thyroid nodules, resulting from a careful and systematic literature selection and quality appraisal. Different types of clinical data of various levels of evidence were considerately presented. Nonetheless, this review is generally prone to inaccuracies from low study quality, study heterogeneity and different types of bias. For some of the assessed diagnostics, the limited number of available publications and small study cohorts contribute to heterogeneity of data and loss of applicability. This mainly concerns studies on non-routine imaging techniques. By nature, these clinical studies need to prospectively include subjects to voluntarily undergo an extra investigation with – at least in the clinical validation phase – no implications for individual patient management. These types of studies require more resources than ‘further use’ tissue biobank studies. Consequently, the number of studies is more limited and published series often are small. In contrast, cytological biomarker research gratefully profits from available large tissue biobanks for initial validation studies. We believe consistent results from properly designed imaging studies should not be disregarded due to mere their sample size, but be appreciated by the quality of their study design and statistics. Population-level study differences were often observed, not only related to test performance but also strongly varying malignancy rates that were oftentimes much lower or higher than expected
RkJQdWJsaXNoZXIy MTk4NDMw