Information for Clinicians: Commercially Available Molecular Diagnosis Testing in the Evaluation of Thyroid Nodule Fine-Needle Aspiration Specimens

Overview

F ine-needle aspiration (FNA) cytology remains the best-established diagnostic test for evaluation of thyroid nodules. However, 10% to 40% of FNA specimens are cytologically indeterminate. Such diagnostic ambiguity often requires repeat FNA, and in some cases, the need for diagnostic lobectomy for definitive histologic diagnosis. Molecular diagnosis is an evolving field that is increasingly being used as an adjunct to cytology to improve diagnosis and prognosis, and better inform treatment decisions for many types of cancer. Molecular diagnosis is also rapidly emerging as a useful tool for evaluation of cytologically indeterminate thyroid nodules that would otherwise require diagnostic surgical biopsy (1). Several molecular genetic tests are now available and offer claim of improved diagnostic sensitivity and specificity. Two commercially available tests that are used in conjunction with FNA include the Veracyte Afirma ^® gene classifier and Asuragen miRInform™ thyroid panel. Both approach molecular diagnosis from different methodological perspectives. The Veracyte method is intended as a rule-out test; that is, the analysis is most effective in identifying lesions with indeterminate cytology that are highly likely to be benign, therefore obviating the need for diagnostic surgery. The Asuragen panel is a rule-in method that is designed to have a high predictive specificity; that is, a positive test is highly associated with malignant histology. A third test developed and offered by the Cleveland Clinic uses quantitative reverse transcription–polymerase chain reaction (RT-PCR) to detect circulating thyroid cancer cells by assessing serum concentration of thyrotropin receptor (TSHR) mRNA from a standard peripheral blood specimen.

Despite differences in approach, the developers of all three methods claim that they improve diagnostic accuracy beyond that obtainable with traditional FNA cytology. Great attention is being focused on identifying how these methods best fit with established practices for diagnosis and treatment of thyroid nodules. The intent of this statement is to briefly summarize these methods and provide direction for clinicians and patients regarding the current state of thyroid molecular diagnosis.

This statement was approved by the American Thyroid Association (ATA) Board of Directors and officially endorsed by the ATA Clinical Affairs Committee.

The Veracyte Afirma Gene Classifier

The Veracyte Afirma method is a multigene expression classifier that assesses gene expression from mRNA isolated from needle washings during a standard FNA procedure. Gene expression is currently compared against 167 genes Veracyte has previously identified as characteristic of the genetic signatures of benign and malignant nodules using a proprietary algorithm. The analysis seeks to determine whether the nodule is benign or suspicious (2,3). A small, published validation trial with 24 indeterminate FNA specimens, using an earlier version of the Afirma gene classifier, showed a 96% negative predictive value (NPV) for identifying histologically benign thyroid nodules (3). Another abstract reporting a similar validation trial with 43 indeterminate specimens using the Afirma gene classifier reported a 96% NPV (2). Recently, Alexander et al. (4), reporting results of the largest clinical validation study to date, showed that for 265 cytologically indeterminate nodules, an overall NPV was 93% with a sensitivity of 92% (i.e., 7 of 85 cancers, 8%, were incorrectly identified as benign). Specificity in this study was 52% (i.e., 48% of benign nodules were incorrectly identified as suspicious). The data also demonstrated that 10%, 10%, and 6% of cancers are missed for nodules with indeterminate cytology in the atypia/follicular lesion of undetermined significance (AUS/FLUS), follicular or Hürthle cell neoplasm/suspicious for follicular neoplasm (FN/SFN), and suspicious for malignant cell (SMC) categories, respectively (Table 1). Alexander et al. (4) ascribe these missed cancers largely to inadequate cellularity of specimens submitted for GEC testing subsequent to cytopathology evaluation. All of these published validation trials (2 –4) have been supported by Veracyte, which has marketed their gene classifier as a rule-out test; that is, the analysis is most effective identifying which lesions with indeterminate cytology are highly likely to be benign, therefore obviating the need for diagnostic surgical biopsy.

A projected five-year cost-effectiveness analysis of the Veracyte Afirma gene classifier method has been published by Li et al. (5). This analysis evaluates cost of treatment for a hypothetical cohort of adult patients with thyroid nodules using a Markov decision model based on 2009 ATA guidelines for management of patients with thyroid nodules (6). The analysis assumes a retail test cost of $3200, which represents the current level of reimbursement offered by the Center for Medicare and Medicaid Services. The analysis shows that the cost of care with current standard of care practice without molecular testing was $12,172 per patient and decreased to $10,719 when the Veracyte test was included. The analysis showed these cost savings were primarily due to the number of unnecessary diagnostic surgeries avoided when the Veracyte test classified a nodule as benign. In this cost analysis study by Li et al. (5), published in 2011, test sensitivity was assumed to be 91%. The cost analysis states that only 1.4% of malignant nodules would be missed by the classifier, similar to the number of malignant nodules missed by the current standard-of-care methods.

According to Veracyte, the Afirma gene classifier is currently not covered by some insurance plans. However, Veracyte offers a plan to defray testing costs for patients who do not have an insurance benefit that covers the Afirma gene classifier test as well as for those who can show financial need.

The Asuragen miRInform Molecular Panel

The panel of molecular tests offered by Asuragen is also intended for use with FNA specimens in conjunction with standard cytology. The molecular tests included in this panel have been previously acknowledged by the ATA, which recommended on the basis of expert opinion that they may be considered for patients with indeterminate cytology on FNA to help guide management (6). This panel identifies mutations in several key cell-signaling intermediaries previously identified to have strong associations with papillary and follicular thyroid cancers. These mutations include point mutations of BRAF and RAS, as well as rearrangements of RET/PTC and PAX8/PPARγ. Although Asuragen has made this panel commercially available, they have not independently published validation of this method. However, there is a published experience with the noncommercial use of these markers as prospective predictors of thyroid cancer in FNA specimens from several academic medical centers (7 –9). Multiple publications have now verified that one of these nonoverlapping (i.e., mutually exclusive) mutations can be identified in well-differentiated papillary and follicular thyroid carcinomas in up to 75% and 70% of cases, respectively (7,8). A recent study reported molecular testing results in 1056 consecutive thyroid nodule FNAs with indeterminate cytology and included an analysis of molecular test and histologic correlation for 513 aspirated nodules (7). This study demonstrates that for all categories of indeterminate cytology, including AUS/FLUS, FN/SFN, and SMC, the risk of malignancy based on cytology only was 14%, 27%, and 54%, respectively. However, if any of the mutations in the panel was identified in the specimen, the risk of malignancy increased to 88%, 87%, and 95% for AUS/FLUS, FN/SFN, and SMC, respectively. The authors state that these dramatically increased risks provide compelling indications for consideration of up-front total thyroidectomy rather than standard diagnostic lobectomy for molecular-positive, but cytologically indeterminate, FNAs. This method demonstrated an overall specificity of 98%. For each of the indeterminate categories of AUS/FLUS, FN/SFN, and SMC, specificity was 99%, 97%, and 96% respectively (i.e., only 1%, 3%, and 4% of benign nodules were positive for a genetic marker). However, overall sensitivity of this method is only 60%, and it fails to identify 37%, 43%, and 32% of cancers in the AUS/FLUS, FN/SFN, and SMC categories, respectively (Table 1). Using this rule-in panel, initial diagnostic lobectomy is still indicated with mutation-negative specimens, but may be eliminated from the treatment algorithm in favor of up-front total thyroidectomy in mutation-positive cases.

Despite the fact that diagnostic lobectomy is still frequently necessary, a recent analysis has demonstrated the cost efficacy of this approach (10). This analysis examines the cost associated with use of the BRAF, RAS, RET/PTC, and PAX8/PPARγ panel for evaluation of nodules with indeterminate cytology classified as either AUS/FLUS or FL/SFN. Cytology SMC was not included in this analysis. The analysis found that when the overall cost of the panel was $650, and molecular testing added $104 per patient to the overall cost of nodule evaluation (standard of care $578 v. molecular testing $682). In this distributed risk model, molecular testing was associated with a decrease in the number of necessary diagnostic lobectomies (9.7% v. standard of care 11.6%) while initial total thyroidectomy was more frequent (18.2% v. standard of care 16.1%). In this model, molecular testing added a diagnostic cost of $5,031 to the cost of each additional indicated total thyroidectomy bringing the total cost to $16,414. However, the cumulative cost was still less than the comparable cost of performing lobectomy ($7,684) followed by completion thyroidectomy ($11,954) in the standard-of-care cohort when indicated by histologic results. In sensitivity analysis, savings were demonstrated if molecular testing cost was less than $870. Asuragen currently plans to offer the miRInform Thyroid panel at a retail price of $2250. Medicare reimbursement for this test is currently $650, while the range of reimbursements from private insurers varies up to $950. The panel would therefore appear to be cost effective where the actual test cost (reimbursement) is below the threshold of $870 used in the noted cost analysis. Asuragen also offers a patient assistance program to defray out of pocket cost.

The Cleveland Clinic TSHR mRNA Assay

This assay may be useful as a marker of thyroid cancer both in the setting of initial diagnosis as well as for disease recurrence (11). When this blood test was performed by itself without FNA in 368 patients with surgical correlation, it was demonstrated to have moderate sensitivity and specificity of 61% and 83%, respectively (Table 1). When only the set of 54 nodules with a cytologic diagnosis of FN/SFN was considered, sensitivity and specificity improved to 76% and 96%, respectively. This study's authors also noted that the further addition of high-quality neck ultrasonography to the diagnostic algorithm incrementally improved sensitivity to 97% with a decline in specificity to 84%. However, these latter statistics would not be applicable to specimens sent to the Cleveland Clinic for evaluation unless the patient was also evaluated at that facility with his or her ultrasonography protocol. These data are based on the single-institutional experience of the Cleveland Clinic and have not been reproduced by others to date. The test is available through direct arrangement with the Cleveland Clinic's clinical laboratory. The Medicare reimbursement for this test is ∼$300, and to date, a cost efficacy analysis for this method is not available.

Summary

TSHR mRNA RT-PCR, the Veracyte and Asuragen commercial methods, and the noncommercial use of BRAF, RAS, RET/PTC, and PAX8/PPARγ testing have promising roles in the diagnosis and treatment of patients with nodular thyroid disease and thyroid cancer. At this time, experience with these molecular methods remains limited, and no test has perfect sensitivity and specificity. However, peer-reviewed data evaluating the diagnostic performance of these tests are increasingly available. The ATA feels that until expert consensus review of existing data (now underway by the ATA Guidelines Task Force) can be completed, no evidence-based recommendation for or against the use of these methods can be made. Clinicians are therefore advised to consider the use of these genetic diagnosis methods with appropriate caution, and to remain cognizant of the limitations of the data supporting their use. Patients who are interested in the use of these tests in their own care should discuss them thoroughly with their care providers. Until evidence-based recommendations are available, determining whether or not the limited data available support the use of these methods should be considered on a case-by-case basis.