Letter to the Editor: Use of Molecular Diagnostic Tests in Thyroid Nodules with Hürthle Cell-Dominant Cytology

Dear Editor:

Thyroid nodules are extremely common and are known to be present in 30–70% of the general population (1,2). While 90–95% of these are benign, 5–10% represent malignancy (3). Fine-needle aspiration (FNA) has been widely implemented to stratify cancer risk of thyroid nodules by using the Bethesda System for Reporting Thyroid Cytopathology (TBSRTC). Approximately 3–25% of thyroid nodules will be classified as cytologically indeterminate (cyto-I) (4). These carry an “indeterminate” risk of malignancy ranging from 6% to 48% with an average of 15–25% (4,5). Consequently, their management has been a matter of debate. Recent advances in cancer genomics have introduced molecular testing of these nodules, which has succeeded in avoiding more aggressive management measures such as diagnostic thyroidectomy (6 –11).

The majority of guidelines in the United States now recommend consideration of molecular testing in patients with nodules exhibiting indeterminate cytology (5), except for those lesions wherein Hürthle cells (HCs), otherwise known as oncocytes, are predominant (12). The HC-predominant FNAs are frequently placed in the Bethesda III or IV categories. HCs are large oxyphilic polygonal cells that contain abundant mitochondria and prominent nucleoli. These cells are present in various benign conditions such as Hashimoto's thyroiditis and benign nodular goiter, but they also dominate the cytologic picture in certain clonal neoplasms such as HC adenoma and Hürthle cell carcinoma (HCC) (13,14). While uncommon, HCC with extensive angioinvasion is known to be one of the more aggressive thyroid cancers with frequent regional and distant metastasis, strong 18F-fluorodeoxyglucose positron emission tomography avidity and resistance to radioactive iodine treatment (13). It has been demonstrated that the majority of HCC do not carry a detectable somatic driver mutation, making accurate characterization with molecular testing that is primarily focused on genomic point mutations and fusions particularly challenging (15 –17). Earlier iterations of molecular testing primarily focused on high sensitivity over specificity to maximize the identification of malignancy in HC-dominant specimens (18). The trade-off was a low specificity and low benign call rate (BCR) (Table 1) (19). Owing to these diagnostic limitations, molecular testing in the early era for HC-dominant cytologic specimens did not decrease unnecessary surgical interventions to the same extent that it did with other cytology categories.

Recently, further details of genomic alterations in HCC have been defined. These changes include widespread losses of heterozygosity/chromosomes and prominent mitochondrial DNA mutations, a genetic landscape that is distinct from other differentiated thyroid cancers (14 –16). Currently available molecular tests, such as the Afirma gene sequencing classifier (GSC) and ThyroSeq v3.0, have incorporated testing methodology that evaluates chromosomal loss and mitochondrial DNA (17,20) and we are now observing an improvement of test performance in real-world studies. We conducted a comprehensive online literature search of the PubMed/MEDLINE, Embase, and Scopus databases that revealed 4 independent postvalidation studies for Afirma GSC (7 –10,21) (Table 1) and 2 independent studies for ThyroSeq v3.0 (17,22), although 1 has combined results with the prior version (Thyroseq v2.0) (17) (Table 2). These studies preliminarily reveal an improvement in specificity and positive predictive value, while maintaining high sensitivity and negative predictive value. Even more accurate assessments of test sensitivity and specificity will require the surrogate endpoint of long-term neck ultrasound follow-up in patients who do not undergo surgery to further reinforce the results from surgical pathology.

Another observation made is an increase in the BCR (Afirma only) or mutation-negative call rate (ThyroSeq only) (23). The BCR is the percentage of molecular tests that yield benign results. For molecular testing with a high negative predictive value, the BCR reflects the fraction of patients who would be treated as if the cytology were benign (Bethesda II). As the majority of molecularly benign nodules typically do not undergo surgery, BCR is directly correlated with the reduced need for diagnostic surgery. The combined BCR among published experiences with HC cytology for Afirma GSC is 68% (Table 3) and that for ThyroSeq is 59% (Table 3).

In summary, recent postvalidation clinical experience studies have demonstrated the improved performance of molecular testing in cyto-I nodules with HC changes (7 –10,17,22). In the available case series, ∼60% of HC specimens report a molecularly benign result that translates to an avoidance of diagnostic thyroid surgery. Although longer follow-up of these cohorts is needed, a reappraisal of the role of molecular testing among HC lesions is worth consideration.