February 2011
Feature Story

William Check, PhD

“A Guide to Diagnostics in Thyroid Surgical Pathology”—a straightforward enough name for a pathology course. But that was only the second half of the title of the session that Jennifer L. Hunt, MD, and Peter M. Sadow, MD, PhD, presented last fall at the 2010 ASCP annual meeting. “Overcoming ‘Thyrophobia’” was the first part of the title—a tongue-in-cheek way of acknowledging the diagnostic dilemmas in thyroid pathology. “A lot of people in practice find thyroid specimens to be very difficult,” Dr. Hunt, associate chief of anatomic and molecular pathology at Massachusetts General Hospital and associate professor of pathology at Harvard Medical School, told CAP TODAY. “Most people I know don’t like them much.”

To do thyroid pathology, in addition to learning the principles of the morphology, it may also be necessary to overcome the psychological hurdle. Drs. Hunt and Sadow addressed both. They identified some of the factors that make thyroid pathology difficult and provided histologic guidelines for interpreting thyroid specimens.

“Thyroid pathology is bread-and-butter pathology,” Dr. Hunt says. “And, honestly, most thyroid cases are straightforward. The hard cases are hard. And they are difficult for people in general practice and for subspecialists alike.” For both groups, one of the most common solutions to hard cases is to consult colleagues. “Show the difficult cases around,” Dr. Hunt advises. “That’s what we do in our subspecialty practice.”

Diagnostic ambiguity is one cause of thyrophobia, Dr. Hunt reports. “There is controversy in the literature about minimal diagnostic criteria. And that makes people nervous.” She was referring to several publications showing high interobserver variability for certain types of thyroid tumors. Despite specializing in thyroid pathology, Dr. Hunt admits these studies make her nervous also: “I have the same controversial cases come across my microscope in my clinical practice.”

Thyroid pathology is challenging for other reasons: lack of diagnostic assays and a potential treatment-prognosis disconnect—the fact that all thyroid cancers, even indolent disease, are often treated aggressively.

Lurking in the background, too, with every thyroid tumor is the possibility of de­differentiation. “A very small number of cases of differentiated tumors will transform into aggressive and usually lethal malignancies,” Dr. Hunt says. “Even though most cancers of the thyroid are indolent, there is always the fear of anaplastic transformation.” Unfortunately, it’s not possible prospectively to identify a tumor that will go on to dedifferentiate and become aggressive, Dr. Hunt says. “That’s one of the reasons why we make sure to identify low-grade tumors and a key reason these tumors are often treated with surgery, even if the great majority of them have very few clinical ramifications for the patient.”

Dr. Sadow, who does ear, nose, and throat, endocrine, and genitourinary pathology at MGH and is an assistant professor of pathology at Harvard Medical School, addressed a second problematic aspect of thyroid pathology—the frozen section. By and large, he advises—and Dr. Hunt concurs—avoiding frozens whenever possible. There’s a cartoon circulating on the Internet, he says, in which a pathologist asks a surgeon why he wants a frozen on a thyroid. “Because I like to tell the patient the diagnosis right after surgery,” the surgeon replies. “That is not a good reason for a frozen section,” Dr. Sadow said.

As part of the gross examination of the thyroid for a frozen section, Dr. Sadow advised finding out the patient’s clinical history, including results from thyroid function tests and autoantibody assays, history of radiation exposure, results of prior FNA, single or multiple nodules, family history of thyroid disease and of heritable tumor syndromes, and drug use, such as amiodarone or minocycline. These data form the context for interpretation of the tissue.

Standard procedure consists of weighing and palpating the specimen, inking margins, and describing any orientation from the surgeon. When processing a thyroidectomy or lobectomy specimen, describe each lesion and evaluate the adjacent soft tissue. Which nodule had a previous FNA? Is there an area of calcification or infiltration? Generally, lobes should be sectioned, superior to inferior, in approximately 3 mm sections to ensure adequate viewing of the tissue. If a frozen section must be done, decide which nodule(s) to freeze and then carefully document which nodule you are freezing for correlation with the permanent section. “Always perform intraoperative smears as well,” he said.

“Freezing causes changes and alterations in the cytology that may lead to misdiagnosis,” he said. In a touch or smear prep, the nuclear cytology is very well preserved and can be used to look for papillary carcinoma.

When is it helpful to do a frozen section? An atypical or suspicious FNA finding is the highest-yield reason to do a frozen section, because some of these lesions can be diagnosed as papillary carcinoma and the patient can thus have a complete thyroidectomy. Secondary reasons might be to assess for parathyroid tissue or for metastatic disease at the time of primary surgery.

The common reason: The surgeon always gets a frozen section. This “indication” heads Dr. Sadow’s list of reasons not to get an intraoperative frozen section. Recently, Dr. Sadow asked a surgeon why he was requesting a frozen section.

“You already know the patient has papillary cancer from the fine-needle aspirate,” Dr. Sadow said. “Why are you asking me to do a frozen?” The surgeon replied, “I just want to know.”

Dr. Sadow’s reasoning: “Unless it alters surgery in some way, why get it? I am going to do a surgical section anyway.” Second on the list of why not to get an intraoperative frozen section is for diagnosis of follicular variant of papillary thyroid carcinoma. “This can be done, but it is difficult,” he said.

An intrinsic reason to avoid thyroid frozens is that freezing disrupts the tissue architecture and capsule, especially for follicular-patterned lesions, and reconstruction may be difficult, resulting in a suboptimal evaluation. This makes a frozen section of the thyroid detrimental and a poor use of resources. Putting fresh tissue into formalin preserves tissue structure so it doesn’t fall apart when you cut it, Dr. Sadow said.

“A frozen section gives you one look at the tumor,” he noted in an interview. “You can’t evaluate the whole thing; it’s just a snapshot. It won’t give you the answer, unless it is a papillary carcinoma. I discourage frozen sections.”

A frozen section is marginally better than an FNA because you have access to the entire tissue. But the advantage is small. Sensitivity of a frozen for papillary thyroid carcinoma is 87 percent, compared with 74 percent for FNA; for follicular variant of PTC the advantage is even smaller—29 percent versus 25 percent (Lin HS, et al. Laryngoscope. 2000;110: 1431–1436; Letuertre E, et al. Am J Clin Pathol. 2001;115:370–374).

In summary, if the FNA is suspicious or atypical, a frozen section can be helpful, Dr. Sadow said. Avoid frozens for FNA with PTC, follicular lesions, or Hurthle cells, or for gross nodular goiter. “It won’t give you any more information,” he said. “Defer if necessary.” One possible wording: “Follicular neoplasm. Defer further diagnosis pending analysis of permanent sections.”

“You don’t need to be a hero and then be wrong,” Dr. Sadow said.

When grossing a specimen for permanent histology, it is important to submit the entire capsule to assess for capsular and vascular invasion in single-nodule, nonpapillary follicular lesions, Dr. Sadow said. If the patient has a previous diagnosis of papillary thyroid carcinoma, or a non-follicular neoplasm on FNA, you can submit representative sections of tumor without submitting the entire capsule. While working with thyroid tissue, Dr. Sadow says, “Always wear personal protective equipment to avoid being splashed with bodily fluids. Cutting into a thyroid, especially when it’s cystic, can be explosive.”

Dr. Hunt discussed four questions pathologists frequently ask about thyroid pathology: Is this an adenoma or a hyperplastic nodule? Is there invasion? Is this a poorly differentiated tumor? Are these features enough for a diagnosis of follicular variant of papillary carcinoma?

In answer to the first question, she says, “It doesn’t matter—don’t waste too much time trying to decide.” Adenomas and hyperplastic nodules are treated in the same way. Identifying carcinoma is the most important goal. Hyperplastic nodules often occur in a background of multinodular goiter. Clinically, nodular goiter can present as multinodular thyroid or with a dominant nodule, usually in a euthyroid patient. Dominant hyperplastic nodules are typically not encapsulated and have the same histology inside and outside the nodule. “Even though most of the nodules in goiter are just hyperplastic nodules,” Dr. Hunt said, “if there is a dominant nodule, it is better to refer to it specifically in your report to avoid a phone call later.”

Diagnosing invasion on thyroid follicular tumors can also be a diagnostic struggle. Dr. Hunt described a three-tiered classification for follicular carcinoma that, she said, is supported by good evidence and gaining ground among specialists and generalists. The traditional approach was to classify follicular tumors as minimally invasive versus widely invasive. “But, my mentor, Virginia LiVolsi [professor of pathology at the University of Pennsylvania] taught me to split out the minimally invasive category into tumors with capsular invasion alone and tumors with angio-invasion.” In one study, five-year mortality correlated with the presence of vascular invasion, rising from two percent with capsular invasion only to 20 percent with vascular invasion and 62 percent for widely invasive tumors (D’Avanzo A, et al. Cancer. 2004;100: 1123–1129). “So at least you should indicate the presence of vascular invasion in your report,” Dr. Hunt said. An unpublished survey by Olga K. Kolman, MD, and Drs. Hunt and Sadow found that among 165 pathologists, about 30 percent of general pathologists and more than 60 percent of ENT/endocrine subspecialty pathologists use this three-tiered classification.

The diagnosis of capsular invasion requires seeing tumor extending through the tumor capsule and blossoming on the other side—what Dr. LiVolsi has described as the “mushroom” appearance, Dr. Hunt said. “If you see tumor definitely outside the capsule, it is most likely capsular invasion, even if you don’t see the point where it goes through the capsule,” she said. Avoid areas with inflammation, degenerative changes, or FNA artifacts, Dr. Hunt cautions. For the diagnosis of vascular invasion, there should be some reaction to tumor within the vessel; for instance, the tumor may be endothelialized, have fibrin on its surface, and be attached to the wall of the vessel. Both vascular and capsular invasion are often best seen on low power. “Low power can be your best friend,” is how Dr. Hunt puts it.

Poorly differentiated thyroid cancer is very important to recognize but a relatively uncommon finding. “It is occasionally difficult to know when something should be bumped up to a poorly differentiated carcinoma,” Dr. Hunt said. The growth pattern of poorly differentiated tumors is often trabecular, insular, or solid. Other features can include necrosis, extensive vascular invasion, marked nuclear atypia, and mitotic activity. If there is a component of poorly differentiated tumor in an otherwise well-differentiated papillary or follicular carcinoma, prognosis is affected; some studies have shown that the typical 80 percent to 90 percent five- to 10-year survival in well-differentiated tumors goes down to 70 percent or even as low as 50 percent, depending on how extensive the poorly differentiated regions are.

Tumors with a differential diagnosis including follicular variant of papillary carcinoma, or FVPTC, make up about 40 percent of Dr. Hunt’s consults, indicating just how challenging these tumors can be in practice. Follicular variant of papillary carcinoma has pure follicular architecture and papillary carcinoma nuclei. Some FVPTCs are encapsulated, while others have an invasive growth pattern. “In follicular neoplasms, distinguishing follicular variant from adenoma is based on one feature: the appearance of the nuclei,” Dr. Hunt said.

Step No. 1 in making the diagnosis is to look at the tumor on low power. “Even at 4∞, you should get a hint that the nuclei are not normal,” she said. “The nuclei say, Look at me on high power.” Clustered areas with nuclear atypia should be apparent at this power and may be more pronounced under the capsule. Compared with normal nuclei from uninvolved background thyroid follicles on the same slide, FVPTC nuclei may have altered nuclear membranes, including flattened or pushed-in sides, irregular contours, chromatin pushed to the edges, and small peripheral nucleoli. Subtle abnormal histological features of FVPTC architecture include perifollicular fibrosis, thick colloid, and abortive papillae.

To supplement these nuclear and histological distinctions and for added certainty, Dr. Hunt said, “people want to use additional diagnostic assays. But perfect magical tools just don’t exist.” It is possible to do special studies with immunohistochemistry and molecular assays. However, these assays may not be very helpful for FVPTC. IHC tends to work pretty well in classical or conventional PTC, but not as well in FVPTC, Dr. Hunt said, adding, “We do get IHC stains on controversial or challenging cases, but most of the time we don’t find the stains that useful.” Possible IHC markers that have been reported in the literature have included HBME-1, galectin, and cytokeratin 19. All are found frequently in conventional PTC and FVPTC, but there is overlapping staining with benign tumors and reactive conditions.

Molecular studies are not much better for FVPTC, she said. Mutations in the BRAF gene are present in about half of classical PTC but much less prevalent in FVPTC. (Most studies report less than 10 percent.) An older marker—RET-PTC translocations—is found in about 35 percent of classical PTC but is also uncommon in FVPTC.

“Because molecular assays and IHC are not usually very helpful,” Dr. Hunt said, “we have to rely on a high-quality H&E for the diagnosis of FVPTC.” Among the pitfalls to avoid when diagnosing suspected FVPTC by histology are FNA artifacts and degenerative changes. “Don’t make a diagnosis of FVPTC just based on an area adjacent to reactive changes or an FNA site,” she said. Finally, avoid nuclear features that are “just too good.” “If every nucleus has a clear inclusion, you should consider the possibility of an artifact,” she said. To determine if this has happened, look at normal cells on the same slide to see if they also have this nuclear vacuolization.

In the question-and-answer period, a pathologist asked Dr. Hunt why specimens that have a very low, almost negligible, potential to metastasize should be called carcinoma. “People always ask that,” she said in an interview with CAP TODAY. “There is no right answer. We could change the terminology and use something like ‘atypical adenoma or tumor of uncertain malignant potential.’” Once she has identified a case that is suspicious for FVPTC, Dr. Hunt does not use these terms: “Even if it doesn’t fit neatly into a benign or malignant category, all treatment protocols require this distinction. We have to make a diagnostic decision.”

The real answer, she says: Less aggressive therapy, to take into account the excellent prognosis of well-differentiated thyroid malignancies, instead of having a one-size-fits-all approach.” Current standard therapy is the same for all follicular derived thyroid carcinomas—thyroidectomy followed by radioactive iodine. But this may be changing. “Some surgeons and endocrinologists are taking a new approach and starting to think more about conservative management of patients who have tumors with very low malignant potential. For minimally invasive follicular carcinoma, it is probably reasonable to do only a thyroid lobectomy and perhaps not treat with radioactive iodine.” After that the patient would be monitored, a kind of watchful-waiting management.

Dr. Sadow discussed several other variants of papillary thyroid carcinoma. While basic features such as extracapsular extension and vascular invasion are the most powerful prognostic indicators in PTC, uncommon variants of PTC can also have prognostic implications. “In a case caught early on, a variant can explain somewhat why it has metastasized,” Dr. Sadow says. Some variants are much more likely to metastasize. Papillary thyroid carcinoma as a rule has an excellent prognosis, and 95 percent of patients survive. The five percent of those who don’t survive tend to have one of these variants.

“Some variants can indicate a concomitant neoplastic process,” Dr. Sadow said. His first example was the case of a 19-year-old woman diagnosed with PTC that was identified as the cribriform morular variant. This variant’s unique histologic appearance and reactivity with thyroglobulin, TTF-1, and beta-catenin help to distinguish it. More important, cribriform morular variant can indicate other tumors and the presence of familial adenomatous polyposis, or FAP. “Cribriform morular variant is associated with FAP about 25 percent of the time,” Dr. Sadow said. “When you find this variant, look for colon cancer as well.” In the young woman, colonoscopy and biopsy revealed multiple tubular adenomas with focal high-grade dysplasia.

Two examples of other variants that Dr. Sadow described and illustrated are as follows:

Diffuse sclerosing variant: This variant is rare, occurs in children and young adults, and is often more aggressive than conventional PTC with lymph node metastases and frequent distant metastases. It has conventional papillary architecture and PTC nuclear features, but has squamoid changes and abundant psammoma bodies, a dense lymphocytic and sclerotic background, and extensive lymphatic permeation.

Oncocytic variant: This variant is grossly mahogany brown (as in renal oncocytomas) and often presents with lymph node involvement (about 50 percent) and multifocal disease (about 35 percent). While it has conventional papillary architecture, it is oncocytic with focally prominent nucleoli and often has psammoma bodies (about 33 percent).

To recognize variants of PTC, Dr. Sadow suggested simple guidelines: “In young patients, be aware of associated syndromes to point clinicians toward additional diagnoses. These tumors will tip you off, just be aware.”