For years, thyroid nodules have been evaluated using fine needle aspiration (FNA). For most patients, this approach works reasonably well: In about 70 percent of cases, the cytology results clearly point to either the presence (10 percent) or absence (60 percent) of a malignancy, thus offering clear guidance on any further interventions (Curr Opin Oncol. 2012;24(1):35-41).
But what about the roughly 30 percent of thyroid nodules characterized as having “indeterminate” cytology? These nodules, which often contain a follicular growth pattern that has proven resistant to definitive cytological evaluation, are frequently subjected to partial surgical removal and, in more than half of those cases, the pathology results come back benign (Diagn Cytopathol. 2008(6);36:425-437).
“Basically, those patients with benign disease just underwent an unnecessary surgery,” said Yuri Nikiforov, MD, PhD, director of the division of molecular anatomic pathology and of the Molecular Anatomic Pathology Lab at the University of Pittsburgh School of Medicine and the University of Pittsburgh Medical Center (UPMC). “And that’s not the only problem: Patients whose frozen section results are indeterminate but the final results indicate cancer then have to come back for a second surgery to remove the entire thyroid.” Such repeat surgeries result in additional morbidity and higher health care costs, he noted (Ann Surg Oncol. 2004;11(1):94-98; Otolaryngol Head Neck Surg. 2004;131(5):596-600).
Since 2007, Dr. Nikiforov and his colleagues at UPMC have been using a method that significantly reduces this “gray zone” of indeterminate cytology. The method involves testing FNA thyroid samples for a panel of genetic mutations that have been shown in previous studies to correlate strongly with the presence or absence of thyroid cancer (Endocr Relat Cancer. 2007;14(4):1089-1097). If the test results point to benign disease, those initial diagnostic surgeries are avoided, he noted. In contrast, if the results help establish a diagnosis of malignancy, a single, up-front total thyroidectomy is performed, eliminating the two-stage surgical approach.
“Empirically, we knew this strategy was working; we were avoiding many unneeded surgeries,” Dr. Nikiforov said. However, there was a lack of hard data to confirm its efficacy. That began to change three years ago, when Dr. Nikiforov and colleagues from UPMC and two other academic medical centers published the first large prospective study of molecular testing of FNA samples for mutations (J Clin Endocrinol Metab. 2009;94(6):2092-2098). The study showed that in a subset of 52 FNA samples with indeterminate cytology, all 15 nodules that were found to be positive for a mutation were malignant at surgery, for a positive predictive value (PPV) of 100 percent. Among 333 mutation-negative nodules from patients who were followed for several years post-surgery, nine were found to be malignant, for a negative predictive value (NPV) of 97 percent.
The panel includes seven main types of mutations that account for most thyroid cancers, according to Dr. Nikiforov: BRAF, NRAS, HRAS, KRAS, RET/PTC1, RET/PTC3 and PAX8/PPARg. The mutation with the strongest correlation to malignancy was BRAF, he added, pointing to another important benefit of the test: its prognostic value. “Research has shown that BRAF-positive thyroid cancers tend to be very aggressive [J Clin Endocrinol Metab. 2008;93(10):3943:-3949],” he explained. In such cases, he noted, surgeons at his institution will keep a sharp eye out for lymph node involvement and, in some instances, opt for a larger excision.
More Recent Evidence
In 2011, Dr. Nikiforov and his colleagues at UPMC published another study aimed at measuring the diagnostic value of the molecular panel (J Clin Endocrinol Metab. 96(11):3390-3397), but with a tighter focus on patients with FNA samples of indeterminate cytology. A total of 967 FNA samples collected from 729 patients whose nodules were found to fit into one of three categories of indeterminate cytology were subjected to mutational analysis. Of that total, 479 patients underwent thyroidectomy, either because repeat FNA testing had previously indicated an elevated risk for cancer or based on the results of molecular testing. The mutational status of those patients was then correlated with surgical pathology.
The investigators found that a positive test for any of the mutations included in their panel correlated with a surgical finding of malignancy in 87 percent to 97 percent of cases, depending on the type of indeterminate findings present. Conversely, the rate of cancer found in mutation-negative nodules ranged between 6 percent and 28 percent.
“The results confirmed the central benefit of this test: For patients with indeterminate cytology, it enabled us to give them definitive treatment recommendations very early in their evaluations,” Dr. Nikiforov said. “If they are positive for any of the mutations, we recommend total thyroidectomy rather than a two-phase surgical approach. On the other hand, if the nodules are negative for these mutations and belong to the lowest-risk category of FNA cytology, they do not need a partial lobectomy or any other type of diagnostic surgery and can be followed by annual routine thyroid examination.”
Dr. Nikiforov added that both studies had some limitations. In the 2009 study, for example, molecular testing was performed without checking for the adequacy of thyroid nodule material collected for molecular analysis. “This could have lessened the chances for mutation detection,” he said. In the 2011 study, the adequacy issue was addressed by refining the collection process, but other potential biases were present, including the fact that the pathologists were sometimes aware of the results of molecular analysis.
Cost-Effectiveness Data in Press
An even more recent study co-authored by Dr. Nikiforov showed that testing for tumor-specific mutations is not only clinically effective; it can be cost effective as well. In the study, which has been published in the Journal of Clinical Endocrinology and Metabolism (2012 Mar 14. [epub ahead of print]), the researchers determined that if the cost of the testing is less than $870 per patient, then savings from reduced interventions offset the cost of the test.
“Because our cost analysis was based on Medicare data, it is widely generalizable to other hospitals that use the molecular testing,” said lead author Linwah Yip, MD, an endocrine surgeon at the University of Pittsburgh. “Using molecular markers to help us determine the most appropriate initial operation can reduce health care costs and, more importantly, not having two operations is safer and more beneficial for patients”
—Linwah Yip, MD
ATA Treatment Guideline
Current treatment guidelines, issued in 2009 by the American Thyroid Association (Thyroid. 2009;19(11):1167-1214), give the molecular testing strategy employed by Dr. Nikiforov and his colleagues a “C” designation—that is, the test is recommended “based on expert opinion.” In contrast, tests and treatments given an “A” rating in the guidelines are “strongly” recommended because they have “well-designed, well-conducted studies in representative populations that directly assess effects on health outcomes.”
Bryan Haugen, MD, a co-author of the ATA guidelines and a co-investigator with Dr. Nikiforov in the 2009 prospective study, said there is probably a bit of a lag behind that “C” rating; evidence is starting to build, he said, showing that molecular testing as an adjunct to thyroid nodule FNA can benefit select patients. Still, “this is a relatively new technology, and we need the results replicated at other institutions,” said Dr. Haugen, who is head of pathology within the division of endocrinology, metabolism and diabetes and Mary Rossick Kern and Jerome H. Kern Chair in Endocrine Neoplasms Research at the University of Colorado School of Medicine in Denver.
Dr. Haugen said that, in his view, the greatest strength of the molecular panel used by Dr. Nikiforov is its PPV. “If surgery is at all a possibility based on cytological findings or ultrasound, then Yuri’s test is the way to go because it can really help guide the type and extent of surgery,” he said. For patients with indeterminate cytological and/or ultrasound findings, in contrast, “you’re in a gray zone, where I agree with Dr. Nikiforov that the risk for doing unnecessary surgery is quite high, because when you take out the nodule, anywhere from 60 percent to 80 percent of them are benign. But, in my view, I think there is a better test available for those scenarios.”
Specifically, Dr. Haugen was referring to Afirma, a proprietary gene expression test marketed by Veracyte. The test, like the one advocated by Dr. Nikiforov, is used to analyze FNA cell samples, but instead of employing a panel of tumor-specific mutations for analysis, Afirma assesses mRNA expression levels for 142 genes that have been shown to be expressed differently in benign and malignant thyroid nodules (J Clin Endocrinol Metab. 2010;95(12):5296-5304). The data are then analyzed with a proprietary algorithm that identifies whether a sample’s expression pattern conforms to these genetic markers for benign tumor growth. The test can only be done at a single lab, located at Veracyte’s headquarters in San Francisco, according to Dr. Haugen.
In September 2010, Dr. Haugen presented data on the test’s efficacy at the 14th International Thyroid Congress in Paris, France. The data were based on a prospective, multicenter study and showed that among 43 indeterminate thyroid FNA samples, the mRNA test had an NPV of greater than 95 percent, according to a press release issued by Veracyte. “A test with this high of [an NPV] will help [physicians] rule out malignancy and confidently monitor many patients with ambiguous thyroid nodules, enabling these patients to avoid unnecessary surgery,” Dr. Haugen stated in the press release.
In his interview with ENT Today, Dr. Haugen confirmed that he is still very much a proponent of the mRNA expression test, adding that the full results of the multicenter study he presented in Paris will be published in a major endocrinology publication “in the next month or two.” But he stressed that each molecular test, the panel of mutations in use at UPMC and the mRNA expression assay “has its strengths, and both in fact are used at our institution.”
Another Vote for Veracyte
Paul Ladenson, MD, director of the division of endocrinology and metabolism, Johns Hopkins University School of Medicine in Baltimore, prefers the mRNA approach. “The strategy of mRNA expression profiling in Veracyte’s Afirma test is to identify benign nodules, not cancers. This is its real strength,” Dr. Ladenson told ENT Today. “Currently, the limitation of any technique that primarily focuses on mutation detection to identify cancers is that at least one in five thyroid cancer patients don’t have any mutation we know about yet and can look for. And so there is going to be an inherent lack of negative predictive value in any test that seeks only known cancer-associated mutations.”
Dr. Nikiforov acknowledged that roughly 30 percent of thyroid cancers are caused by as yet undefined mutations. In the 2009 study, that gap was underscored by the fact that the sensitivity of the panel, when used alone, was only 62 percent, he noted. That’s why he stressed the value of doing the mutational panel in concert with cytologic testing, which can help nail down the pathology of thyroid nodules and yield a more definitive diagnosis. In the 2009 study, he noted, that approach yielded an NPV of between 92.9 percent and 97.4 percent. “And the NPV of this test is only going to get better in the next few years,” he stressed, “when I am confident that the incredibly powerful genetic research tools we currently have will reveal the remaining mutations that cause thyroid cancer.”
A slightly more tempered view of molecular testing was taken by Constantine Theoharis, MD, assistant professor of pathology at Yale University School of Medicine in New Haven. In a review paper focusing on the molecular diagnosis and management of thyroid nodules (Curr Opin Oncol. 2012;24(1):35-41), he challenged the idea that BRAF mutations detected with the panel used by Dr. Nikiforov and others can serve not only as a diagnostic marker for malignancy but also as a prognostic indicator of aggressive disease that requires more extensive surgery. “While there may be some empirical evidence supporting this approach, there are no long-term prospective studies that I’ve seen in the literature that link any of these clinical strategies to BRAF status. Such studies would be exceptionally difficult to conduct,” he said.
Dr. Theoharis said that although the Veracyte assay is not in use at Yale, it is noteworthy. “As far as I understand from [the developer’s] presentations, they’ve taken an alternative approach: Rather than attempting to prove malignancy, they try to prove that a given nodule is benign. That is quite innovative, but I am unclear if it is cost effective,” he said.
mRNA Test Also Proves Cost Effective
Dr. Ladenson said that such cost effectiveness data do, in fact exist: He and his colleagues at Johns Hopkins recently published a study (J Clin Endocrinol Metab. 2011;96(11):E1719-E1726) on the mRNA test’s cost effectiveness and concluded that it reduces the overall financial impact of managing patients with thyroid nodules. The researchers based their analysis on a hypothetical group of adult patients with cytologically indeterminate thyroid nodules. A wide variety of factors that could influence cost were included in the analysis, including those related to inpatient care, primarily surgery, and outpatient care, such as clinic visits and sonography. All calculations were based on 2010 Medicare payment rates, including reimbursement for the test itself, which is set at $3,200. Quality of life (QOL) estimates and the probabilities of pursuing various management options were also included in the analysis.
Their model showed that when current practice patterns were followed—that is, no mRNA testing—57 percent of patients with indeterminate cytology underwent surgery for what ultimately proved to be a benign nodule. When the mRNA test was routinely applied to such patients, only 14 percent had surgery. On a per-patient basis, the cost of care for patients who did not undergo mRNA tests in the model was $12,172, versus $10,719 for those in whom management was guided by the mRNA test.
Those seemingly modest savings are a function of performing “a judiciously designed cost-modeling study,” Dr. Ladenson said. “For example, using Medicare payment rates in our base case, about $3,000 per surgery, biases the analysis against this novel test.” In many real-world clinical situations, he explained, where up to 30 percent of patients are on fee-for-service private insurance plans, “the costs of thyroid surgery can be $10,000 dollars or more.” In such cases, he noted, “You can see how the cost savings from avoided surgeries can really add up.”
Even using the lower-scale payment rates allowed by Medicare, he noted, nationwide savings are impressive. “Our analysis predicted that [due to the mRNA test] 74 percent fewer surgeries would be performed in patients with benign thyroid nodules,” he said. “If the mRNA test was performed in all of the 450,000 patients annually who have indeterminate cytology results, that translates to a savings of $122 million in the U.S.”
Personalized Medicine Gets a Boost
Asked to review the cost-effectiveness data from Dr. Ladenson’s hypothetical patient group, Dr. Theoharis said the findings “are just that, hypothetical. I’d like to see those savings documented in a real-world clinical trial.”
Dr. Yip agreed that the Ladenson et. al cost analysis has some flaws. “To not do surgery based on the Veracyte NPV of 95 percent means that you will miss cancer 5 percent of the time, ” she said. “According to their clinical model, if their test is negative, patients are not followed up at all. In their cost model, they do not address the health, emotional, and legal impact of a missed cancer.”
But such debates should not overshadow the fact that when used in the right patients—that is, in the correct clinical-radiologic context, in conjunction with pre-operative cytomorphology—molecular testing for selected mutations “can serve as an effective quality control tool for the cytopathologist, allowing a valid diagnosis to be made on indeterminate morphology, often with scanty [thyroid nodule] material.”
In doing so, Dr. Theoharis added, these molecular tests “help further the growing field of personalized medicine,” in which a patient’s genetic makeup can help clinicians choose optimal therapy.
Disclosures: Dr. Haugen disclosed that he has received research support from Veracyte through his university but no direct salary support from the company. Dr. Ladenson disclosed that he is a consultant to and received past research support from Veracyte. Drs. Theoharis, Nikiforov and Yip had no relevant financial relationships to disclose.