The Role of Genetic Testing in the Evaluation of Thyroid Nodules - HD

Introduction

Hi, I'm Joe Langer from the Pearlman School of Medicine at the University of Pennsylvania, and I'll be giving you a talk about molecular testing of fine neural aspiration of thyroid nodules.

Role of Molecular Analysis in Thyroid Nodule Evaluation

A common question is, where does the molecular analysis of FNA specimens fit into the evaluation of thyroid nodules? Certainly each of us, from our different disciplines has a different perspective whether we're looking at from the ultrasound point of view, from the point of view of the endocrinologist and risk factors, or from the point of view of the pathologist.

Now, as you may know, each nodule is either a cancer or not a cancer. And thyroid FNA is the best non-surgical diagnostic tool to diagnose malignancy. But thyroid FNA is not always able to distinguish a cancer from a non-cancerous lesion, and only surgical pathology is able to make that distinction.

Types of Thyroid Cancers

When we talk about thyroid cancer, we're primarily talking about epithelial tumors, which are derived from the follicular cells. These account for about 95% of all tumors, the most common cell type, are well differentiated papillary thyroid cancer, which account for 80%, and the pathologist will subdivide these into pure papillary cancer and follicular variant of papillary cancer. This subtype has some micro follicle formation in it, and the pathologist at their discretion will divide between these two different ca types of papillary thyroid cancer.

Less commonly, we have well differentiated follicular cancer, and about 5% are poorly differentiated papillary cancers, and less than 1% anaplastic the most aggressive type of thyroid cancer. And we have medullary thyroid cancers, which are derived from the para follicular C cells, accounting for about three to 5% of all thyroid cancers.

Now, the particular subtypes that may look different from the traditional papillary thyroid cancers are those that are follicular internally. These would be the follicular variant of papillary thyroid cancer and well differentiated follicular cancer. And we'll talk about why that is particularly important as we talk about the different types of FNA results.

Follicular Cancers

Now, pur follicular cancers comprise about 10% of all thyroid cancers in the United States, but the incidence is higher in iodine sufficient areas of the world, such as Europe and Sub-Saharan Africa. These tumors are more common in women than men and have a little bit of an older demographic than papillary thyroid carcinoma with a peak in the sixth decade. Unfortunately, their 10 year survival is not as good as papillary thyroid cancer, about an 87% 10 year survival.

Now, one of the things that's very important to note is that histologically, the benign follicular adenoma is very similar in appearance to the follicular cancer. The only distinguishing feature of the cancer is we can identify capsular invasion or extension into blood vessels on surgical pathology. Of all follicular tumors, 80% are benign adenomas and 20% are thyroid cancers.

So here's an example of a follicular cancer. The central part of this lesion here under the microscope looks very similar to an adenoma, the benign tumor. The pathologist cannot make the distinction by looking at any of the features of these cells. Our pathology colleagues make the distinction by observing capsular invasion of the tumor into the fibrous capsule demonstrated here, or vascular invasion into the blood vessels outside of the lesion. This is the only way to make the diagnosis of follicular cancer. It's not a diagnosis that the pathologist can render based on just the FNA specimen.

FNA Diagnoses

Now, when we do thyroid FNA and we hand our specimen to the pathology colleagues, they can render one of four diagnoses. They can make a diagnosis of a benign nol, and they do so when they see heterogeneous population of follicular cells that have relatively no atypical features, often add mixed with colloid and macrophages.

They can render a malignant diagnosis on the FNA specimen if they see features of papillary tissue fragments, abnormal nuclear features such as irregular contours or nuclear grooves, intra nuclear cytoplasmic conclusions. And these features are highly predictive of an underlying papillary thyroid carcinoma.

They could give us a non-diagnostic diagnosis, and this occurs when we have an insufficient number of cells in the aspirate. The number of cells we actually need to achieve a diagnostic level is very high. When we talk about thyroid FNA, we need to have approximately six to eight groups of cells, each group composed of at least 10 cells.

But the very problematic biopsy result in terms of patient management is the so-called indeterminate category. In this category, the pathologist sees monotonous follicular cells, but without any atypical nuclear features.

What could a indeterminate thyroid biopsy result be? Underlying the lesion could be a benign follicular adenoma. It could be a cancer, either a pure follicular cancer or a follicular variant of papillary cancer. It could even be a benign nodule, but the FNA specimen was just obtained from a region that had relatively monotonous follicular cells. It could even be a focal area of chronic lymphocytic thyroiditis.

Old Schema and Malignancy Risks

Now, until recently, this was the schema that was used for giving prognostication of malignancy risk based on cytology results. At the malignant end, if a biopsy was read as malignant by our pathology colleagues with very high likelihood it would prove to be a malignant lesion at the time of cervical resection, approximately five to 15% of all FNAs fall into this category in most ultrasound labs.

At the other end of the spectrum, the vast majority of biopsies fortunately fall into the benign category, but note that this is not a 100% likely to be benign lesion at the time of pathologic evaluation. I've given you a 96% plus. The range is 97 up to a hundred percent benign in the literature, but you'll notice that it's not 100% uniformly.

Now, as I mentioned before, the criteria for diagnostic accuracy is relatively high with respect to thyroid FNA. And interestingly, a non-diagnostic result is also not a zero risk of cancer. In some series, the risk has been as high as 10%. The non-diagnostic rate will vary depending on the particular expertise of that institution, if the nodule cystic and so forth, and approximates about 10 to 20% of all FNAs.

But the most problematic biopsy result in terms of patient management is this indeterminate category, which occurs in 20 to 30% of all FNAs. And unfortunately, this particular result had between a 15 and a 75% cancer risk, obviously a very broad cancer risk. One that was not very helpful for patient management decisions.

What was the standard of care? When patients would get this result, the majority would go on to have that low of the thyroid removed. The rationale was about 20% of the time this was a cancer, but 80% of the time it was benign that lobe would be removed. And if the patient fell into the malignant category, they would then need to have the other lobe removed. But this led to up to 85% of patients having surgery for what proved to be a benign lesion because of the indeterminate pathology result, surgery was pursued, but the lesion proved to be a benign adenoma or other benign lesion.

And certainly, thyroid surgical surgery does have some risk. It has a risk of recurrent laryngeal nerve injury, as well as permanent hypothyroidism. Some patients would be directly referred to have the whole thyroid out based on an a follicular cytology result. This was often in patients who were considered high risk, by patient preference sometimes, or if there were some features that the pathologist would intubate on the report, that in their opinion carried more suspicion than other features of a follicular neoplasm reading.

And as a pathologist thought about this, they noticed that within this category of follicular neoplasm, there were features that did make them more or less worried about a particular thyroid nodule.

Bethesda Classification

So in 2008, the Bethesda classification was developed. One of my colleagues, Bert Belo at the University of Pennsylvania, it was instrumental in this particular new classification. And in essence, what they did is they took this indeterminate result and made three tiers. The highest level of suspicion within this indeterminate category was now called suspicious for malignancy. And a very lower level of suspicion is now called follicular lesion of uncertain significance or atypia of uncertain significance flusser us. And then there were still nodules that were very similar to the previous category of follicular neoplasm. They could not be downgraded nor upgraded within this category.

But what was very helpful about this category is now we had a very more tightly aligned malignancy risk with each particular diagnosis. As I mentioned before, the non-diagnostic unsatisfactory category does not have a 0% risk, but about a one to 5% risk. The benign cytology, two to 4% atypia of uncertain significance or follicular lesion of uncertain significance, a five to 15% risk follicular neoplasm, 15 to 30% suspicious for malignancy, 60 to 75%, and malignant 97 to nearly 100%.

Management Based on Bethesda Categories

And so now management can very much be dictated by these particular risk categories as before malignant diagnosis on FNA is very predictive of thyroid cancer. And patients can be referred for a total thyroidectomy, as I mentioned before, as well. Benign is not 100% benign, but these patients can be safely observed with follow-up sonography.

And those who have a non-diagnostic result on the first FNA, it's recommended they undergo a second financial aspiration. And fortunately in the vast majority of them, the second aspiration in combination with the first will achieve cellular adequacy, and the patient can be given a definitive diagnostic category.

We're now gonna turn our attention to the three new categories created by the Bethesda Classification. And these were the risks that we just discussed. What should one do? If you get a diagnosis of flu on the first biopsy attempt, the recommendation is to repeat this biopsy. When you repeat the biopsy, at least 50% of the nodules will now be considered benign at the time of the repeat biopsy. And these patients can be managed as though they received a benign diagnosis at the first go round.

A little bit paraphrasing from what the pathology colleagues have discussed with me about this particular category is that the first biopsy had some very mild atypical features, not enough for the pathologist to be worried. They were dealing with a malignancy, but not quite enough for them to call this a benign aspirate at the first sitting when they repeat the biopsy. And now there's a broader sample or more cells are now observed under the microscope, they can reassure themselves that there's random atypia and not a more sinister pattern. They feel very comfortable rendering a benign diagnosis. And as I mentioned before, these patients will go into an observation group.

Now, about 15% of the time, an upgrade can occur with the second biopsy, not showing random ayia, but more suspicious ayia or more suspicious nuclear features than the first biopsy. In this case, the pathologist feels as though there may have been an under sampling of the suspicious area of this nodule. And now the patients are treated as though they've been rendered a suspicious diagnosis for papillary thyroid carcinoma and become a surgical candidate.

And about 35% of the time, the patients will remain in a relatively indetermined category, either a diagnosis of follicular neoplasm on the second biopsy, or still at the plus US level, and will talk more in depth about how to handle these types of patients. So of the patients that get a second flosser us, the risk is likely higher than the first group. The 50% of patients that were benign have now come out of the sample. And so we now quote a patient a higher risk, but we may not have a firm number. Most of these patients, therefore, will have the option of either surgery or observation. A little bit tempered by clinical factors, as before we have patients with a diagnosis of follicular neoplasm. 20% of these will be a cancer and 80% will be benign.

Should we offer these patients hemithyroidectomy, just as before, should we offer them total thyroidectomy of the patients that are suspicious for malignancy? Most institutions will do a total thyroidectomy, but some patients may say, gee, if I only have a 40 percent chance that my nodule is benign, I would be comfortable with a hemithyroidectomy and then repeat surgery if I turn out to be one of the ones that have a malignant potential.

So this is as far as diagnostic cytology can go on these aspirate and a number of investigators, were looking to see if any of these risk stratification could be improved by methods other than just fine neo aspiration and observation of the cellular characteristics under the microscope.

Molecular Testing for FNA Aspirates

So two different types of testings for the FNA aspirate have been developed. One actually looks at genetic mutations within the cells and the second looks at the mRNA or the gene pool within the individual cells found within thyroid nodules. And we're gonna talk about these two different types of tests individually.

The goal of these tests is very different. The goal in patients who have a who are already going to the operating room because they have a relatively high risk of cancer, we would wanna have a test that gives us even a firmer diagnosis, a high predictive value that in fact, that nodule is almost certainly to be a cancer. And those individuals, then all of those patients would be referred to for a complete thyroidectomy, and you would not offer the option of a hemithyroidectomy because of a higher predictor of malignancy.

At the opposite end of the spectrum, if you have a nodule, for example, that has a follicular reading of undetermined significance or a follicular neoplasm, and it's a relatively nons suspicious appearing nodule, perhaps there's a test that could give them even lower risk of cancer than based on just the FNA risk. And so this is a different type of goal. This goal is a higher likelihood of a benign diagnosis to try to avoid surgery for those patients who would opt for sonographic surveillance.

Oncogene Testing

So let's talk about the first type of test. The oncogene testing. The genetics of thyroid cancer has to a reasonable degree, been very worked out. And two different types of chromosomal abnormalities have been identified. The first are point mutations, and many of you may be familiar with the BRAF mutation, which is a mutation seen in all sorts of cancers, and that has a role in thyroid cancer, as does the RAs mutation. The other are chromosomal rearrangements, and the RET PTC, for example, has been very well worked out as well when these can be identified in cells with relatively high predictive value. We know we're dealing with a thyroid cancer, and as I'll demonstrate just in a moment, BRAF, perhaps one of the best predictors.

The one of the most extensively studied mutations is the B-M-A-B-R-A-F mutation. And this very nice meta-analysis that was published in 2011 looked at three over 3000 nodules that had been studied to date. And you'll note that in the benign category, there was only one positive test out of this entire category of benign nodules that expressed this particular mutation. And this was thought by many to be a lab error.

On the other side, you'll notice that papillary cancers, about 45% of them will express the BRAF mutation. But in the poorly differentiated anaplastic cancers, only 20% will show the BRAF mutation and follicular cancers having a very different histology. None of these will express the BRAF mutation.

So the genetics of thyroid cancer, as I mentioned before, have been very well worked out. The BRAF mutation, in fact, is a predictor of actually a more aggressive type of papillary thyroid cancer when it's found, it's found in some of the more aggressive tall cell variants, and it's also linked with extra thyroid extension. It also tracks with a higher stage of presentation. So the BRAF gene, when it's noted to be positive, is a predictor of an underlying papillary thyroid carcinoma. But as I mentioned before, it's not present in all of the cancers.

The RE PTC is a less common genetic mutation, and this is linked with radiation exposure. It's actually seen in patients that are younger and it's seen with the classic histology. So although these patients are diagnosed at a younger age, they tend to have a better prognosis.

The follicular variant of papillary cancer typically does not have the BRAF or the re PTC mutation, but instead is more likely to either have the RAs or the P-A-X-P-P-A-R gamma type of genetic defect. And this is much more similar as we'll see to the follicular lesions. So again, follicular variant of papillary cancer, it is a papillary cancer, but it tends to be much more similar to follicular cancer and dissimilar to the more classic types of papillary thyroid cancer.

Now, there were three papers which in fact used mutation panels to analyze patients with these FNA diagnoses. Before they were referred for surgery the NICO four F paper looked at follicular neoplasm, suspicious for follicular neoplasm and the flu category, and the QUIN and the Moses paper only looked at follicular neoplasm and suspicious for malignancy, the CONT and the Moses looked at these genetic rearrangements. And the Nico Phora group also considered this. And you'll notice that this was a very high specificity test. That is when a particular nodule tested positive for the panel of gene mutation with very high likelihood it proved to be a thyroid cancer. So very useful in particular for counseling patients who had follicular neoplasm and plus, including the PPAR gamma before operation. Had you known that the nodule was positive for the genetic panel with very high likelihood a cancer was underlying this particular lesion.

Now, the problem with these tests is they're not optimally sensitive, and that means that not all cancers express them. In fact, in this particular paper here by Dr. Moses of follicular neoplasms that prove to be cancer, less than 30% in fact were gene positive, meaning that many cancers are not gene positive. So you could not use this test to exclude a cancer. You can only say that when a test is gene positive, it's a highly predictive of an underlying cancer. And additionally, there were some false positives. There were no false positives when we looked at these genetic panels, but there were false positives for the RAS, and these were the false positives that were reported in the literature and in the Moses series, up to 40% of the nodules.

And if you look at the genetics of follicular adenoma and follicular cancers, we can see why. And that is that both adenomas and follicular cancers do share some genetic rearrangements that overlap. And so, again, this is problematic that we cannot distinguish between cancers and adenomas based on these particular types of genetic rearrangements.

The RET PTC and the BRAF avir are usually going to be positive in papillary thyroid cancers. But these type of follicular lesions remain the challenge, and unfortunately, again, have a significant overlap even in the genetic expression of these particular lesions.

So the mutation panels, again, you would not use these to avoid surgery, meaning a negative test will not cancel surgery, but they're very helpful in a patient who is going to the operating room. If you're trying to decide if you should do a hemithyroidectomy or total thyroidectomy in that, if you are gene positive, particularly with the BRAF mutation, you almost certainly have an underlying carcinoma.

Now, they are not as helpful in the follicular neoplasm and plus, because in these particular categories, many of the cancers, in fact, will be oncogene panel negative, particularly if you're not using this particular gene in your panel, that's suspicious for papillary cancers will tend to be gene positive. So again, it's a test that has some utility in some very specific clinical applications.

How do we use it? We do use it when we have a suspicious for malignancy and the patient is trying to decide between hemithyroidectomy or complete thyroidectomy. We tend not to use it in most patients with plus or follicular neoplasm, but we do use it if we feel as though watchful waiting is not likely to be an option to be offered to that particular patient. Those are patients where we think that the nodules risk of malignancy is probably higher than a typical nodule in these categories.

What would make us think that patients have a higher risk than the whole category would be if they had a large nodule, if they had suspicious sonographic features or something about the demographics of that patient, such as if they were of male gender. If they had a solitary calcified nodule, or for example, let's say the nodule was detected by PET scanning, these would be more suspicious background features and we would probably just recommend the patient have a complete thyroidectomy.

Now, there is some work that has used this particular test in the non-diagnostic category, again, noting that not all of these nodules are benign and up to 10% can be malignant. And a nice paper by Canterra, when they applied it to nodules that had indeterminate inadequate cytology, in essence, they were able to identify 12 cancers on the basis of positive oncogene panel testing.

RNA Expression Tests

We're gonna switch gears, and we're now gonna talk about RNA expression tests. And this is a relatively novel technique, which is looking at the genes within thyroid nodule aspirate, and looking at the genetic makeup of these particular nodules, what proteins are expressed by the mRNA within those particular nodules. And this was a very interesting way of looking at thyroid nodules.

In essence, it was a test desire to have very high negative predictive value. That means if you could look at the genetic makeup of a particular cell, and you could say, this cell is not expressing any genes that have been linked to thyroid cancer, perhaps then we can say this is a particular type of thyroid nodule that has a much lower risk than based on its classification of plus or follicular neoplasm by, its just looking at the cytologic features.

And the way that this was developed was to take aspiration of cells and take it to the lab and use mRNA technology to do a micro, a microarray analysis of all the genes in a particular cell, and then to use multidimensional algorithmic mathematical modeling to, in essence distinguish between genes that were and were not important in the setting of thyroid cancer versus benign thyroid nodules.

And here's the schematic of how this works. So within any particular cell, we have 22,000 genes. And it's thought that about 3000 of these genes are actually expressed. Many are turned off, many are turned on. And ultimately there was about 142 genes that were identified that were thought to be relevant in patients with thyroid cancer. This was then further attested against a group of known thyroid cancer, and they realized about a hundred of these 142, they were missing about 25 genes in the group of known thyroid cancer, such as medullary cancer. One was a case of malignancy metastatic to the thyroid, and they added in 24, 25 more genes to come out with the total 167 genes, which now comprise the assay.

This work was published in the New England Journal of Medicine. Two of the lead authors in this particular paper were from my institution, and it was a very nice, well performed 19 month long prospective multicenter validation trial involving all different types of sites, both academic private practice with a very large number of patients, much like you would use this in the clinical setting.

And here were the results. In particular, what they found that in nodules that either had a tipia or flus on the second biopsy and those with follicular neoplasm diagnosis, that there was a very high negative predictive value. That is to say that of the aspirate that went on to have this mRNA analysis, if they lacked any of those 167 genes in the panel, none of them were expressed with very high negative predictive value. They proved not to be a malignancy.

If you look here, it's suspicious for malignancy, the negative predictive value was 85%. So that is still a 15% likelihood of malignancy, not really low enough in many people's eyes to obviate the need for further evaluation.

And so now if we can go back to the chart about management, we can see that before we know if we had a benign diagnosis in hand from cytology, the risk of cancer was two to 4%. If we now have a patient who's flu times two and has a negative gene classifier or follicular neoplasm and a negative gene classifier in someone we would consider a low risk patient, the risk now can be quoted to that patient to about five to 6%. Not that different from non-diagnostic, not that different from benign and under certain well-defined parameters. Those patients may in fact be a group that we can offer observation with ultrasound follow up, much like we follow our patients with benign cytology.

And this was now released for commercial use. Then in the first six to nine months that it was used, 51 endocrinologists were surveyed and they had 368 patients in whom they used the gene classifier expression test. And you can see what happened. They had a very dramatic shift in patients that were, they were referring to surgery from 74 patients down to eight patients. So a large group of patients were, did not need to have surgery. The vast majority proving to have benign a surgical pathology, but instead, we're in the watchful waiting group. And so this was actually very helpful and the endocrinologist felt very comfortable offering this to their patients.

Now, just like with any new test, when it's released from its experimental phase into widespread use, you may get different results. And these were results from the Mayo Clinic. In their experience using the test outside of the testing phase, they were a test site, but in practical application, and these were the results of a little over a thousand patients, you can see that they had about 8% of their population, or 89 patients who had the flus times two or follicular neoplasm. And the patients were offered genetic testing.

They had 53 who opted into have the genetic testing, and they had 30 who were not submitted either because the patient declined or the physician thought that the nodule was suspicious, or the patient's demographics were such that they did not feel that watchful waiting would be a useful option such that even a negative result would not obviate the need for surgery. And they had six patients in whom they had insufficient sample. And at when the test was first launched, many patients were only having one aspirate. We now routinely prefer two aspirate in order to have enough material.

So ultimately, 53 patients went on to have the genetic testing. And here's what happened. They had 13 or 24% of their patients that had the gene classifier in the benign range, meaning that no abnormal genes were expressed. And these patients underwent watchful waiting. They had 40 patients or nearly three quarters of their population that had the suspicious result. 13 had no surgery, 20% went on to have surgery, and the vast majority of them proved to have benign nodules.

So remember that it's a negative predictive value test. If you fall into this group, we can say with very high negative predictive value, you have a benign nodule. The converse is not true. It does not mean if you express any of these genes that you're highly likely to have a malignancy. Remember, several nodules such as follicular adenoma may express some genes that overlap with thyroid cancer.

Well, this result of 2023 patients, or 85% benign and 15% cancer, was much less than was in the New England Journal article, which quoted a 38% malignancy rate. What's different about these two types of studies? Well, one is you have to remember that it all hinges on the cytologic diagnosis. If you have pathologists who are more likely to put a patient in follicular lesion of undetermined significance versus a benign category, then clearly you're gonna have more benign biopsies, excuse me, more benign surgical pathology. And so a lot of this may have to do with how your cytopathologists are classifying the nodules. And then you may just have different patient population in terms of your risk of cancer.

And it's very important to note that the particular negative predictive value that occurred in this New England Journal medicine article was specific to that patient population. If you change the prevalence of disease, if you change anything about the reading of your cytologist, you in fact may have very different results. So you have to be very careful about taking the results that were in the New England Journal of Medicine and blanketly applying this to any particular risk. Everything about the particular negative predictive value and positive predictive value. It's very fixed for a prevalence of malignancy and the sensitivity and specificity in that particular cohort.

So here you can achieve 94% negative predictive value, but let's say that you have a very high prevalence of malignancy in your particular cohort. You can see you would have exactly opposite results. So you have to be a little bit careful about using these particular results without thinking about your particular patient population.

Additionally, the negative particular value is going to fall as the disease prevalence increases. And you can see that in these ROC curves that this is the 95 sensitivity and this is the 80% sensitivity and that the negative predictive value in fact, is quite lower even at the same prevalence of disease. So you have to be very careful about just applying these.

Practical Use of Molecular Tests

So how do we use these in practice? Well, if we have a diagnosis of follicular lesion, of uncertain significance or a follicular neoplasm and we think that the patient has just the generic risk of cancer, there's nothing about the demographics of these patients such as a male with a solitary nodule, there's nothing about the ultrasound appearance of this nodule. There's no additional clinical history, for example, a positive PET scan, then we will in fact apply the genetic expression classifier. We know it has relatively high negative predictive value in our particular patient population. And with the benign result in hand, we will offer that patient the observation arm. They will undergo sonographic surveillance at approximately yearly intervals.

If it is suspicious, we are off a lobectomy knowing that the vast majority of these patients will still have benign disease. Remember, we will not offer this if the patient does not agree to go into the observation category or if there's some other risk factor. If the nodule is large, if it has suspicious sonographic features, we're not going to use observation in those particular patients because a benign result is not going to achieve a very low negative predictive value.

Summary

So in summary, there's two different types of tests out there, oncogene testing, which identifies genes expressed in thyroid cancer. And these has a very, very high positive predictive value, but relatively low sensitivity. This is a rule in test. We use this when we know the patient is already going to go to surgery, and in fact we wanna know is it a definite cancer? And we would offer the patient a complete thyroidectomy differently that the gene classifier expression is a test that has very high negative predictive value. We use this when observation is an option for that patient when we don't want to have to send the patient to surgery to prove that a nodule with very high likelihood is benign. And we tend to use this in plus times two small follicular neoplasms and those with no suspicious sonographic features.

So in 2014, we start just as always with history physical TSH, we use ultrasound for risk stratification to decide which nodules need an FNA, we perform the FNA and we get the cytology results. If the results are benign, malignant or non-diagnostic, we have a management algorithm with non-diagnostic biopsies having a repeat biopsy, we can consider the mutation panel, the oncogene testing in this particular group, if we are rendered an indeterminate lesion, follicular lesion of undetermined significance, follicular or suspicious for malignancy, we have two options. We can do molecular analysis to further triage the risk of cancer. We report the result and then refer the patient for surgery if necessary.

And going back 50 years to a New England journal article from 1964, I think we still have the same dilemma. How to select a patient with nodular thyroid gland in whom there's a greatest likelihood of finding a cancer is still subject of much disagreement, but now in 2014, we can offer molecular analysis and molecular analysis will continue to evolve. There's a lot of work done now in improving those oncogene panels. We now have identified both false negative and false positives and oncogene panels with 2030 different oncogenes are about to be released commercially. And so you can look forward to a further refinement of this particular type of molecular analysis when we have indeterminate cytology.

Thank you very much for your attention.