An emerging market of molecular markers for thyroid cancer diagnostic and prognostic evaluation challenges current medical and surgical management; however, here is a strong word of caution; Unless the markers have been tested in the context of clinical and surgical management their true efficacy has not been accurately assessed.
We must always ask two questions before ordering these tests that can cost more than $3,000: 1) Does the use of a molecular test add any additional information that would otherwise change our recommendations; in other words, is the clinical information already known enough for appropriate clinical decision making? 2) What is the basis upon which recommendations for its use have been made; in other words, how were the studies conducted and interpreted? Are they valid?
By way of background, the gene expression clarifier (GEC) Afirma is marketed as a diagnostic tool for the differential diagnosis of indeterminate thyroid cytology (atypical cells of undetermined significance or follicular neoplasm, Bethesda categories III and IV, respectively).
Most importantly, it has been tested and is marketed as a negative predictor only. It is also only useful if used in the appropriate setting and if the test result is negative; a positive Afirma test can tell us only that the nodule is indeterminate, something, by definition, already determined by fine needle aspiration (FNA). Furthermore, any study examining its efficacy must be conducted within the context of clinical scenarios. For example, for a patient with an indeterminate lesion and a symptomatic multinodular goiter, neither a negative nor positive Afirma result would have any impact on surgical decision making and yet would have cost the patient more than $3,000. The latter patient simply needs a thyroidectomy. A patient who has a nodule that is approaching 4 cm or is increasing in size or a patient who has undergone several FNAs, ultrasounds, and endocrinology consultations similarly would likely choose surgery regardless of the GEC test result.
The only scenario in which Afirma testing is useful is in a patient with otherwise no indications for surgery except an indeterminate FNA; for instance for a patient with an asymptomatic < 4cm, stable, single nodule. In this scenario only a negative result could be helpful and would obviate the need for surgical intervention. One also must ask what is the likelihood of the patient undergoing repeat FNA, having repeat indeterminate cytology and repeat Afirma testing over the ensuing years and at what point should we consider these costs prohibitive compared with the patient undergoing a thyroidectomy instead?
Similar concerns exist with regard to somatic mutation panels, the most commonly marketed and used being Asuragen. The mutation panel consists of Ras and BRAF mutation and RET/PTC and PAX8/PPAR-gamma chromosomal rearrangement analyses. Importantly, the expanding literature on molecular markers for thyroid cancer, however, has lost track of a vast body of literature that reports Ras mutations present in up to 30% of benign thyroid nodules; RET/PTC in 24% and PAX8/PPAR-gamma in 14%, dramatically challenging the validity of the panel purported to accurately identify thyroid malignancy. We know that BRAF mutation is exclusively found in papillary thyroid cancer, but generally a BRAF-positive tumor is one in which a cytological diagnosis is already clear; it is either suspicious or definitively malignant on FNA. One might suggest that these aforementioned markers might indicate a harbinger of malignancy or identify a precancerous lesion; however, there is no study that supports this theory. Furthermore, it is important to note that even among pathology experts the diagnoses of follicular adenoma, follicular cancer, and follicular variant of papillary thyroid cancer have significant and documented inter- and intra- observer variability, thus challenging the gold standard upon which these markers have been tested.
With regard to BRAF, Ras, RET/PTC and PAX8/PPAR-gamma as prognostic markers there are many considerations, the least of which includes the differential diagnosis of follicular lesions addressed above and the known presence of these abnormalities in benign tumors. The majority of studies to date have conducted only univariate analyses without consideration of the clinical variables already available to us to make needed clinical decisions. In order to examine the usefulness of any marker panel a multivariable study is required, incorporating the available clinical variables such as size, lymph node metastases, and extra-thyroidal extension. For instance, the majority of studies examining the presence of lymph node metastases in BRAF-negative vs. BRAF-positive tumors were almost uniformly conducted on patients in whom routine lymph node dissection was not performed, thereby skewing the results. Only patients with suspicious nodes underwent a lymph node dissection; thus for those patients who had no suspicious nodes there was no information about their nodal status. Furthermore, several studies that included patients who underwent routine lymph node dissections and, importantly, included patients who underwent prophylactic dissections for whom a marker would be most useful, did not find that BRAF mutation was associated with aggressive features, including the presence of lymph node metastases. The same can be stated for a panel that includes Ras, RET/PTC, & PAX8/PPAR-gamma. “Follicular variant of papillary thyroid cancer and follicular cancer” will more likely harbor Ras and PAX8/PPAR-gamma mutations and will, by definition, include some tumors that would be considered benign by other pathologists.