Increased molecular testing accelerates precision thyroid nodule management, cancer care
Thyroid nodules are common. Each year, approximately 600,000 U.S. residents with nodules undergo a fine-needle aspiration biopsy in which cells are extracted and examined to determine whether the nodule is benign or cancerous.
Fine-needle aspiration, or FNA, biopsy is mostly accurate and most nodules are benign. However, in approximately 20% to 25% of cases, the test returns an indeterminate finding, meaning there was not a conclusive identification of benign or malignant thyroid disease. Only a decade ago, most of these patients would go on to diagnostic surgery, with approximately 60% overtreated or undertreated, based on the surgery that they receive.
“Ten or 15 years ago, when someone had indeterminate cytology for their nodule, there were really only two options — to watch the nodule conservatively or to surgically remove it,” Erik K. Alexander, MD, chief of the thyroid section at Brigham and Women’s Hospital and professor of medicine at Harvard Medical School, told Endocrine Today. “The problem, of course, is most patients do not want to watch things when there is a potential risk for cancer, so we were doing a lot more surgery, much of it unnecessary and much of it for benign disease.”
Today, management of indeterminate thyroid nodules has expanded from a repeat FNA biopsy or diagnostic surgery to include molecular testing that measures genomic changes in tissue samples taken from the thyroid. Such tests are part of an effort to improve preoperative risk assessment and reduce unnecessary surgeries, which come with their own risks. Since validation data for the first widely used molecular test was published in 2012, molecular analysis of thyroid tissue has become a generally relied upon tool to help triage patients with indeterminate cytology.
“The rationale for developing these tests was really about getting a high negative predictive value and high sensitivity so that you could exclude patients from surgery; we now call that the benign call rate,” Susan J. Mandel, MD, MPH, chief of the division of endocrinology, diabetes and metabolism, professor of medicine and radiology and co-director and creator of the Thyroid Nodule Clinic at the University of Pennsylvania, told Endocrine Today. “[The test] would catch the cancers — yes, it may give you some false positives — but it was not going to miss cancers.”
Currently available molecular tests are useful for ruling out malignancy in indeterminate thyroid nodules; however, longer-term follow-up studies are needed to confirm that negative nodules are truly benign, and some researchers suggest discrepancies exist between study and real-world data on predictive values. Further study is also needed on analysis of specific gene mutations to help identify aggressive disease to guide prognostication and management.
“The molecular tests are increasingly found to be cost-effective,” Alexander said. “They are certainly preventing unnecessary surgeries, which carry risk. Now we are trying to intelligently determine how to use molecular medicine, with an understanding that it is value-added — but we need to figure out in which ways it is value-added.”
Beyond the microscope
The Bethesda system for reporting thyroid cytology classifies nodules that undergo FNA into six groups of varying malignant potential: I (nondiagnostic), II (benign), III (atypia or follicular lesion of undetermined significance), IV (follicular neoplasm), V (suspicious for malignancy) and VI (malignant). In cases of benign cytology, the American Thyroid Association recommends periodic monitoring with imaging. For cases of malignancy or suspicion for malignancy, the recommendation is typically to undergo thyroid surgery, either lobectomy or total thyroidectomy.
The gray area for endocrinologists and endocrine surgeons is nodules that fall into Bethesda categories III and IV. Previously, recommended next steps were a repeat FNA or a diagnostic surgery.
“These indeterminate nodules have puzzled thyroidologists for decades because most nodules were still benign on final histology,” Christian Nasr, MD, a staff physician in the department of endocrinology at Cleveland Clinic’s main campus, wrote for the ATA’s Thyroid Health blog in September.
That uncertainty — and the large number of patients who ultimately underwent an unnecessary surgery — helped drive the development of molecular methods that would analyze the FNA samples “beyond just looking under the microscope,” Nasr wrote.
“Before the era of molecular markers, people don’t realize that one of three patients went to surgery,” Stephanie L. Lee, MD, PhD, ECNU, associate professor of medicine, director of thyroid health in the section of endocrinology, diabetes and nutrition at Boston Medical Center and an Endocrine Today Editorial Board Member, said in an interview. “Anyone who had a malignant biopsy, and anyone with an indeterminate biopsy, which in some places may be anywhere from 25% to 35%, depending on your cytopathologist, went to surgery. What that gave you was a really high sensitivity. It means you did not miss many cancers. The specificity was terrible. For every person with a cancer, we sent four or five people without cancer to surgery.”
There are several diagnostic tests on the market for application to indeterminate thyroid nodules. The first molecular tests were the ThyroSeq v2 genomic classifier, which evaluates DNA mutations within a 19-gene panel, and Afirma gene expression classifier (GEC), which evaluates messenger RNA expression patterns for 167 genes. Initially, costs for the available molecular tests were high, and negative predictive values were not good enough to warrant wide adoption.
“When molecular markers came out, I actually did not start using them right away,” Lee said. “In part, because they were too new; I didn’t know what they really meant.”
Both tests have been updated since their initial release to utilize next-generation sequencing. ThyroSeq v3 now offers next-generation sequencing that evaluates for point mutations, gene fusions, copy number alterations and abnormal gene expression in 112 thyroid cancer-related genes. The Afirma genomic sequencing classifier (GSC) is based on RNA sequencing technology evaluating 10,196 genes.
Additional commercial tests have included assays of oncogenic microRNA expression, such as ThyraMIR and Rosetta GX, which is no longer in clinical use, and a seven-gene panel test for oncogenic point mutations and gene fusions.
“In our current practice, with the development of molecular testing over the last decade or so, we are at the point where the negative predictive value associated with many of these tests is sufficiently high that we can manage patients with negative testing as if they had benign cytology,” Sarah Mayson, MD, associate professor of endocrinology, metabolism and diabetes at the University of Colorado School of Medicine, told Endocrine Today.
The Afirma GSC test will provide the clinician with a binary result — positive, which is suspicious for cancer, or negative, which is benign. A separate add-on test, the Afirma Xpression Atlas, provides information on 761 DNA variants and 130 RNA fusion partners in more than 500 genes that are associated with thyroid cancer.
“The purpose of Xpression Atlas is to identify genetic alterations that could potentially affect prognosis or management of an individual patient,” Mayson said. “If a person has advanced thyroid cancer, for example, it could identify a potential therapeutic target. That test itself is also RNA-based, so it has some limitations in terms of what it can detect.”
For ThyroSeq, genomic alterations are the basis of the test; a clinician will automatically receive the information about what genetic alteration is identified along with the allelic frequency.
“ThyroSeq is different because it looks at everything — DNA and RNA, gene expression variants and copy number alterations,” Mandel said. “You can be positive for multiple different things. Based upon what you are positive for, the test gives you a sense of what that tumor is going to be. If you are positive, for example, for a RAS mutation alone, the likelihood of cancer or [noninvasive follicular thyroid neoplasm with papillary-like nuclear features] is 60% to 65%. But if you are positive for RAS and another mutation, your likelihood of cancer is 90% to 100%. If you are positive for a copy number alteration alone, the likelihood of cancer is only 50% to 60%. You can begin to understand what that tumor is going to look like and then how that tumor will behave.”
In an analysis published in JAMA Oncology in December, Masha Livhits, MD, FACS, assistant professor of surgery in the section of endocrine surgery at UCLA, and colleagues conducted the first study comparing diagnostic performance of the two current molecular test techniques. The researchers assessed data from 346 patients in the UCLA Health System who underwent thyroid biopsy from August 2017 to January 2020 with indeterminate cytology for Bethesda III or IV nodules, randomly assigned to Afirma GSC or ThyroSeq v3. Primary outcome was diagnostic test performance of the RNA test (Afirma) compared with the DNA-RNA test (ThyroSeq).
Researchers found that both the RNA test and DNA-RNA test displayed high specificity. Diagnostic thyroidectomy was avoided for 51% of patients tested with the RNA test and 49% patients tested with the DNA-RNA test. Surveillance ultrasonography was available for 90 nodules, of which 94% remained stable during a median 12 months of follow-up.
“Although previous trials demonstrated that the prior version of the DNA-RNA test was more specific than the prior version of the RNA test, the current molecular test techniques have no statistically significant difference in performance,” the researchers wrote.
More data needed
Studies point to improved diagnostic certainty with the most widely used molecular tests that can help guide treatment; however, some experts suggest data supporting the routine use of molecular testing are limited.
In a review published in Frontiers in Endocrinology in October, Martha A. Zeiger, MD, an endocrine surgeon with the surgical oncology program at the National Cancer Institute, wrote that newer versions of ThyroSeq and Afirma have addressed past issues of lower real-world positive and negative predictive values, respectively, than initially reported. ThyroSeq v3, for example, has been shown to have a negative predictive value of 97%, with benign call rates of 61% for indeterminate nodules and 53% for nodules with Hürthle cells. The newer Afirma GSC has a higher specificity while retaining a similar negative predictive value compared with the Afirma GEC. However, discrepancies between initial reports and experience from real-world use are already appearing with the newer-generation tests, she wrote.
“As was learned in the case of Afirma GEC and ThyroSeq v2, these initial findings and test performances need to be confirmed in both independent studies and in larger cohorts,” Zeiger wrote. “Until then, caution should be urged regarding these tests in achieving both diagnostic certainty and truly impacting clinical care in patients with indeterminate thyroid nodules.”
Zeiger also noted the lack of any study documenting an effect and appropriate change on surgical management and argued that real-world studies “undermine the cost-effectiveness touted by previous modeling studies.”
“Like any diagnostic test, these expensive investigations are best used in a thoughtful manner and in a case-by-case basis for nodules with equivocal clinical, cytopathologic and radiographic factors as recommended by professional societies such as the American Thyroid Association and the American Association of Endocrine Surgeons,” Zeiger wrote. “Molecular testing should only be considered if the result of a test would otherwise alter the recommended management of the patient.”
The newest molecular tests provide much more genomic information compared with previous versions; however, endocrinologists caution that the data must be interpreted carefully.
“One might think if ThyroSeq is using identified mutations, you should be able to figure out if [a nodule] is positive or not,” Lee said. “The reality is they expanded their panel to include more than 100 genes. What they are now starting to pick up are genes with a lower risk of association with cancer. Having a positive ThyroSeq does not mean a positive test for cancer anymore.”
The lack of clarity has been disappointing for some endocrinologists, Lee said.
“One of the disappointments for practicing endocrinologists is they thought these tests would provide a yes/no answer. It’s not,” Lee said. “With Afirma, you can use it as a yes/no, but now they added on the gene expression atlas, which makes it like the ThyroSeq. And then ThyroSeq added on gene expression abnormalities, so they are becoming more like Afirma. Each of them are competitive and are trying to merge into the same technique. It does not change the fact that they are very good at ruling out cancer and are not so good at being able to say that every lesion is benign.”
Alexander, who led the clinical validation study for the original Afirma test, said available molecular tests are now able to solve the problem of unnecessary surgery.
“All the tests are aligned with their metrics to maximize their negative predictive value,” Alexander said. “I do think for these leading tests, when you get a benign result, you should be confident that that really is a benign nodule. In that sense, they are accurate. Where they are different is that no single test can tell you conclusively whether every nodule is either benign or malignant.”
The use of molecular markers has been limited by insurance companies — molecular testing for a Bethesda V or VI nodule, for example, is not covered, as a negative result cannot prevent surgery. However, Lee said the use of presurgical molecular profiling of malignant nodules should be seriously considered.
“Take the example of the person with multiple mutations or knowing it’s a BRAF mutation up front. Right now, we are no longer able to do these molecular profiles on a Bethesda V or VI, which are going to be cancers,” Lee said. “Not all cancers are the same. That is a huge question: Can using the molecular markers for people who ultimately have cancer define what the appropriate management of the cancer should be?”
Mandel said there is potential utility for molecular profiling of Bethesda V nodules when that information could help clinicians better predict “behavior” of the tumor.
“For Bethesda V, the distribution of results may be a classic papillary tumor or a higher risk tumor,” Mandel said. “Could the genetic profile, preoperatively, better predict which patients need bilateral surgery? People are shifting now to use molecular testing in Bethesda V’s to help make that decision.”
Molecular testing is also advancing in real time; researchers and clinicians still do not know what will ultimately make up the final panel of gene expression or gene mutations that are associated with thyroid malignancy.
“We have the low-hanging fruit — easy, yes/no answers,” Lee said. “Now we are starting to look at multiple combinations and finding out the combinations that are important to predict malignancy.”
“There are still genetic alterations that we do not yet have sufficient knowledge of, such as how their presence might affect an individual’s prognosis, especially those alterations that are less common,” Mayson said.
A bigger unanswered question remains for people who have negative molecular testing in the setting of Bethesda III or IV cytology and undergo active surveillance, not surgery.
“One of the challenges is how do you manage a molecular benign nodule,” Mandel said. “That is huge, because that is the majority of the nodules that undergo molecular testing. To me, that is the biggest challenge, and there is very little data. Many of us are using, for example, the baseline ultrasound pattern of the nodule to help decide how to follow them. We need an evidence-based way of managing molecular benign nodules.”
In a recent commentary published in The Journal of Clinical Endocrinology & Metabolism, Alexander wrote that the next step for molecular thyroid care lies in its continued ability to further prognosticate and personalize medical decisions, targeting care only where most needed.
“There is so much still to discover as to the basic underpinnings of how cellular biology and genomic medicine works,” Alexander said. “But it is very unlikely that, at any given time in the future, it is only going to be genomic understanding of disease that answers all questions of clinical care. That will never be the case. In fact, we still need to figure out how we integrate our demographic understandings of patients, our ultrasound findings, our cytology — together with genomic data — and how they are each mutually exclusive and independent of each other.”
- Alexander EK. J Clin Endocrinol Metab. 2021;doi:10.1210/clinem/dgab087.
- Dutta S, et al. J Clin Endocrinol Metab. 2021;doi:10.1210/clinem/dgab030.
- Khan TM, et al. Front Endocrinol. 2020;doi:10.3389/fendo.2020.590128.
- Livhits MJ, et al. JAMA Oncol. 2020;doi:10.1001/jamaoncol.2020.5935.
- Mayson SE, et al. Endocrinol Metab Clin N Am. 2019;doi:10.1016/j.eci.2018.10.004.
- Nasr C. Molecular testing in thyroid nodules: It is all about risk for malignancy. Available at: www.thyroid.org/molecular-testing/. Accessed April 29, 2021.
- For more information:
- Erik K. Alexander, MD, can be reached at firstname.lastname@example.org.
- Stephanie L. Lee, MD, PhD, ECNU, can be reached at email@example.com.
- Susan J. Mandel, MD, MPH, can be reached at firstname.lastname@example.org.
- Sarah Mayson, MD, can be reached at email@example.com.
Click here to read the point/counterpoint.