Reply to can a gene-expression classifier with high negative predictive value solve the indeterminate thyroid fine-needle aspiration dilemma?


Rossi and colleagues aptly identify an important issue in the search for appropriate ancillary tests for indeterminate thyroid fine-needle aspiration (FNA) specimens. Given the current expensive price tag of molecular testing, including the Afirma Thyroid FNA Analysis (Veracyte, South San Francisco, Calif), the use of less-expensive alternatives, possibly including immunocytochemical markers as proposed by Rossi et al, could provide a large cost savings, at least until the time (most likely within the not-too-distant future) that molecular testing becomes more affordable. This is not a trivial point. In the current health care climate, in which cost containment is paramount, balancing the efficacy of a particular testing method with its overall cost versus savings to the system is critical. In agreement with Rossi et al, this is one reason why the first-line response to an indeterminate thyroid FNA diagnosis of atypia of undetermined significance (AUS)/follicular lesion of undetermined significance (FLUS) should be repeat FNA, as supported by the Bethesda System for Reporting Thyroid Cytopathology.[1] A significant aspect omitted by Alexander et al's discussion of the Afirma test is the role of repeat FNA in the decision-making process.[2] Several groups, including ours, have shown that repeat FNA can accurately reclassify greater than 50% of nodules in the AUS/FLUS category as benign by cytologic evaluation alone without the use of any ancillary molecular tests. Thus, for the AUS/FLUS group of specimens, which accounts for the largest percentage of indeterminate thyroid FNA cases, a repeat thyroid FNA appears to be at least as good if not better than the Afirma test for guiding patient management.

Specifically with regard to the application of immunocytochemical studies to indeterminate thyroid FNAs, Rossi et al have cited their work using galectin-3 and HBME-1 to triage these thyroid FNAs into low-risk and high-risk categories.[3] Although the negative predictive value of those markers is only 77%, the use of immunocytochemistry in the study by Rossi et al relies not so much on a definitive classification of the thyroid FNA but on a more general sorting of FNAs into 2 management groups. That being said, there is a steep hurdle to overcome to convince laboratories to use immunocytochemistry for thyroid FNAs, namely demonstrating that such testing is reproducibly accurate. In addition to mixed results using galectin-3 and HBME-1 in individual laboratories, there are conflicting reports about the use of these markers within the literature. Nevertheless, even if immunocytochemistry were to be used to evaluate indeterminate thyroid FNAs, its success as an ancillary test is likely to be short-lived as molecular strategies become both more precise and much more affordable.


The author made no disclosures.

  • William C. Faquin, MD, PhD

  • Department of PathologyMassachusetts General HospitalHarvard Medical SchoolBoston, Massachusetts