Fine Needle Aspiration Biopsy Cytology of Thyroid: A Diagnostic Approach Based on Pattern Recognition

Thyroid FNAB Accuracy

Thyroid FNAB has been recognized as the “most accurate and cost-effective method for evaluating thyroid nodules.”

FNAB has a positive predictive value (PPV) of greater than 90% overall for diagnosing specific thyroid lesions and a diagnostic accuracy for malignancy up to 98%. It is also highly accurate in diagnosing the various types of thyroiditis.

False negatives do occur because of undersampling of lesions, cystic papillary carcinomas, and lesions where there may be dual pathology, such as papillary carcinoma arising in a background of Hashimoto thyroiditis.

Traditionally there have been problems in distinguishing benign follicular or “adenomatous” nodules from “follicular neoplasms,” which include follicular adenomas and low-grade follicular carcinomas, but over time this has become less of a problem. Of course, the differential diagnosis (DD) of follicular adenoma and follicular carcinoma cannot be made on FNAB because this relies on demonstrating vascular invasion or capsular invasion at the margin of the nodule by histopathology. The clinical outcome for a benign follicular nodule and follicular adenoma is virtually identical, and most low-grade follicular carcinomas have a good prognosis.

Thyroid FNAB Technique

The technique and its possible but rare complications should be explained to the patient and the patient consented in an appropriate manner within the hospital or clinic. Warfarin should probably be replaced with heparin in patients medicated for anticoagulation. The thyroid is a very vascular organ, but use of fine needles carries a low risk of hemorrhage, although bleeding back into completely drained cysts does occur and can be minimized by placing firm pressure on the FNAB site after each pass.

Ultrasound of the thyroid can be used to confirm the palpable nodule and other possible target nodules. Lymph nodes should be carefully looked for because FNAB of these nodes at the same time as the thyroid can provide staging information. Although ultrasound is widely used, the cytopathologist can proceed to the FNAB using palpation alone for palpable nodules.

If the procedure is being performed under ultrasound for either a palpable or impalpable nodule, the cytopathologist can perform the FNAB closely with the radiologist or ultrasonographer, or the radiologist can perform the FNAB and an experienced cytotechnologist or cytopathologist can provide rapid on-site evaluation (ROSE) to evaluate adequacy and triage the specimen for possible ancillary studies such as flow cytometry. ROSE is cost-effective, reducing the number of inadequate specimens and the need for repeat FNAB, which involves costly repeat ultrasound and FNAB, as well as providing best practice and cost-effective use of the specimen and minimizing patient discomfort.

The patient can be in a reclining chair or on a bed, but the neck needs to be extended over a low pillow to expose the thyroid. This should be done carefully in the elderly and may be uncomfortable in those with any arthritic neck condition. The patient should not be left supine in this position for long periods of time.

If there are multiple nodules, then those nodules showing any atypical ultrasound features should be selected, or multiple nodules can undergo FNB or a nodule in each lobe can be sampled.

Fine needles of 23, 25, or 27 gauge should be used. Because nodules in the thyroid frequently have cystic degeneration, use of a needle by itself to perform the FNAB may lead to cyst fluid cascading out of the needle. A syringe can be attached to the needle for the FNAB, or an aspiration syringe holder or “gun” can be used, which frees up the other hand to palpate and “hold” the nodule with the fingertips. Alternatively, if the patient is supine, a needle attached to a syringe with the plunger removed can be used to aspirate cysts.

An attempt should be made if cyst fluid flashes immediately back into the needle hub to drain the cyst completely. This can be quite dramatic when viewed under ultrasound or felt under the palpating fingers. Hemorrhage can occur virtually immediately back into the cyst, so firm pressure should be applied to the cyst bed, after drainage. The operator should record on the request form if the cystic lesion is completely drained.

If there is a residual nodule, either by palpation or seen on ultrasound, a needle by itself can be used to rebiopsy the lesion. This technique, which can also be used for the first pass if the ultrasound shows a definite solid nodule, allows fine control of the needle and facilitates the required rapid “woodpecker” passaging of the needle deeply into the nodule.

At least two and up to four passes or needlings of the nodule are recommended to increase accuracy, first with the syringe holder to drain cysts and then with the needle only, that is, the nonaspiration technique of any residual lesion.

When the needle enters the vascular thyroid, blood appears rapidly in the hub. The needle should be withdrawn after one or two more rapid jabs. Unless the hub fills immediately with brown cyst fluid followed by a surge of easily aspirated cyst contents, the needle should not be kept in the nodule. Filling the syringe with 1 or 2 mL of bright red blood produces poor bloody smears with diluted diagnostic tissue, and invariably dwelling for more than 5 to 10 seconds in the thyroid while ultrasound pictures are taken to “prove the needle was in the lesion” leads to fibrin clotting, especially on the second and subsequent passes.

The essence of the good thyroid FNAB is to use a 25- or 27-gauge needle to rapidly stab the nodule, agitating the needle into and out of the nodule rapidly over a 5- to 10-second period, and remove it immediately after blood enters the hub. Aspiration should NOT be used during this “fine needle biopsy.” The exceptions are the cyst that should be drained and the colloid nodule that will resist even aspiration of its tenacious and diagnostic thin colloid.

Both air-dried Giemsa-stained and alcohol-fixed Papanicolaou (Pap)-stained slides should be used for all cases, recognizing that some institutions will have a long tradition of only using one of these stains or a variant such as the H&E stain.

Cell blocks are useful in some circumstances for ancillary studies, such as where medullary carcinoma or metastasis is suspected clinically or at ROSE. It is up to the individual cytopathologist to decide if a cell block should be made.

Liquid-based preparations offer the advantages that all the material is placed in a fluid medium, so direct smears fixed in alcohol for Pap stains will never show “air-drying artifact,” but it is not possible to prepare air-dried Giemsa stains with this technique and, importantly, most of the patterns that provide so much diagnostic information and most of the thin colloid are lost. The notable advantage of the liquid-based preparations is that the whole case is reduced to one slide, and sometimes it helps to reduce inadequacy rates when the smears are pauci-cellular. However, noncytology staff can be trained to achieve good-quality direct smears, and ROSE reduces the inadequacy rates of thyroid FNAB and provides both alcohol-fixed Pap-stained and air-dried, Giemsa-stained slides in a cost-effective manner. The method of air-drying slides at the FNAB and then rehydrating before Pap staining has been advocated.

Thyroid FNAB: Criteria for Adequate Smear

The Bethesda System for Reporting Thyroid Cytopathology (TBS) defines thyroid FNAB adequacy. It requires a minimum of six tissue fragments or sheets. Each sheet has 10 to 20 cells so that the architecture of the sheets can be seen, and a colloid background may be helpful. A single, large tissue fragment of multiple follicles and intervening stroma can suffice.

However, in a colloid nodule with a large amount of thin colloid there may be minimal or no follicular epithelium, but the distinctive appearance of the slide seen macroscopically or when stained is regarded as sufficient by TBS. When correlated with the clinical and imaging findings, these cases should be signed out as a “colloid nodule.” Similarly, with chronic lymphocytic thyroiditis there is no requirement to have follicular sheets in the presence of reactive lymphoid material. If there is any atypia such as intranuclear pseudoinclusions or marked nuclear atypia in any technically insufficient smear, then this atypia should be reported.

Cystic degeneration is common in benign follicular nodules and follicular neoplasms and can occur in papillary carcinomas of the thyroid and in metastases. It is contentious whether follicular epithelium is necessary in the cystic background of siderophages, debris, old blood, and thin colloidal proteinaceous background to make the FNAB adequate. If the lesion is fully drained at ultrasound and the radiologist sees no atypical features on ultrasound, or there is no residual nodule to palpate, it is reasonable to accept a “cyst contents” diagnosis, recording that there was no epithelium. But under the current TBS, cyst contents without follicular epithelium in the absence of ultrasound or clinical information that the cyst has been drained completely is regarded as “nondiagnostic” or “insufficient.”

Thyroid FNAB Request Form

It is essential that the referring clinician provide the standard demographic information, any family history related to thyroid pathology, and history of exposure to radiation and antithyroid treatment to the cytopathologist. Each thyroid nodule undergoing FNAB should be documented as to the size, specific site (lobe and position in the lobe, “upper” or “lower,” etc.), and imaging findings.

Thyroid FNAB Diagnostic Terminologies

In October 2007, the U.S. National Cancer Institute and Papanicolaou Society of Cytopathology sponsored a conference to determine the use of FNAB in the management of thyroid nodules and to standardize the diagnostic terminology, which led to the publication of a classification system and an atlas. The classification included six distinct, well-defined cytomorphologic categories that matched the probability of malignancy at surgery to the category and the corresponding clinical management recommendations ( Tables 4-1 and 4-2 ). It introduced standardized nomenclature and diagnostic criteria and set phrases to be used in reporting diagnoses with the hope to improve intrainstitutional and interinstitutional reproducibility of thyroid FNAB reporting and to gain wider clinical acceptance of thyroid FNAB. It provided relatively unambiguous, clear, and clinically relevant diagnostic categories, linking FNAB diagnosis and patient management. TBS does not advocate using category numbers in reports but rather establishes set diagnostic phrases. Potentially it will replace multiple reporting systems, but it has attracted considerable criticism, as well as support.

The TBS has three categories for indeterminate FNAB because each category has its own risk of malignancy. “Atypia of undetermined significance” (AUS) or “follicular lesion of undetermined significance” (FLUS) are cases that are not diagnostic of a specific lesion and have features that cause some concern but have a low risk of malignancy in the range 5% to 15%. Cases that raise suspicion of a follicular neoplasm, that is, follicular adenoma or follicular carcinoma with capsular or vascular invasion, have a more substantial risk of malignancy, 15% to 30% in the literature, while those that are suspicious of carcinoma, such as papillary carcinoma including the follicular variant of papillary carcinoma, have a high risk of malignancy in subsequent surgery of 60% to 75%.

The AUS/FLUS category is a heterogeneous group, supposedly representing less than 7% of all the FNAB, but more recent publications have shown that the rate of AUS/FLUS can range to greater than 20%. TBS has been criticized for introducing this potential “wastebasket” of cases, which potentially could encourage a lack of rigor in the diagnostic process. TBS sensibly only allows eight specific smear findings for this diagnosis ( Table 4-3 ), but there have been many publications on the institutional impact of the classification system on inadequate rates and AUS rates, using both direct smears and liquid-based preparations. AUS rates have been reported as between 3% and 18% by an analysis of eight papers, and an AUS-to-malignancy ratio at surgery has been proposed as a quality assessment measure, suggesting that a rate of 3:1, AUS to surgically proven malignancy, is appropriate. If the AUS-to-malignancy ratio is greater than 3, then either AUS is being overdiagnosed or malignancy is underdiagnosed.

Recently one of the original authors of the Bethesda classification suggested that the classification should be left as it is for a further period of time, to allow further assessment, before any change is promoted. On the other hand, subclassification of the AUS/FLUS category into those with scant material showing a predominantly microfollicular pattern, prominent oncocytic/Hurthle cells or suboptimal smearing and staining, and those with nuclear atypia suggestive of papillary carcinoma does show that nuclear atypia carries a greater risk of subsequent carcinoma. It is not recommended at present to use “FLUS” for the former and “AUS” for the latter. Applying molecular tests, such as BRAF, which has near 100% specificity for papillary carcinoma, to the AUS/FLUS case with nuclear atypia rather than all AUS/FLUS cases may be cost-effective in determining which patients go to thyroidectomy. However, subclassification of the AUS/FLUS category at the present time has not led to changes in management, and repeat FNAB remains the least expensive and so far proven technique to clarify an initial AUS/FLUS diagnosis. Approximately 50% of AUS diagnoses will show a follicular neoplasm or more significant lesion at repeat FNAB or surgical biopsy.

It should be emphasized that the six categories should never be reported purely as a category number as the diagnosis in a report, but rather the standard phrases embodied in the classification should be used in the FNAB report, to foster clear communication between the clinician and cytopathologist. The categories can be recorded by number for quality assurance statistical analysis.

Ancillary Studies of Thyroid FNAB

Although the Bethesda System for Reporting Thyroid Cytopathology is widely accepted in the United States, the indeterminate cytologic diagnoses pose problems for both pathologists and clinicians. This has led to the development of molecular markers to attempt to improve FNAB diagnosis and tumor prognostication. Over the past decade, a number of molecular markers have emerged as possible diagnostic tools for solving this challenge.

Smear Patterns: Specific Lesions and Their Cytopathological Criteria

When assessing thyroid FNAB smears, an experienced cytopathologist or cytotechnologist can readily make the diagnosis of most cases on the smear pattern using scanning powers, confirming the nuclear features in the epithelial tissue fragments and the nature of the dispersed cells on high power, repeating this process, low to high power, until the findings correlate. If the high power is discrepant, for instance, if plentiful nuclear pseudoinclusions are found in a follicular pattern with colloid, the low power has to be reassessed.

Some disease processes and tumors can present with different patterns. Chronic lymphocytic thyroiditis, which with supporting clinical evidence may be called Hashimoto thyroiditis, can present with multinucleated giant cells and be regarded as “granulomatous” but lacks prominent epithelioid granulomas, while it can also present with mainly lymphoid material and scant epithelium and be regarded as a lymphoid inflammatory pattern. There is a spectrum of lymphocytic thyroiditis and Hashimoto thyroiditis with the cutoff based on clinical and serologic findings and the cytology. The cytoplasm of hyalinizing trabecular tumors can be poorly defined, and it may show syncytial sheets of tumor cells, while on the other hand, it can show more defined cytoplasm and more trabecular tissue fragments. For practical purposes, we placed it in Pattern 7 with papillary carcinoma because the presence of often plentiful nuclear pseudoinclusions leads to a DD with papillary carcinoma.

The term oncocytic rather than oxyphilic or Hurthle or Askanazy to describe the cells in Pattern 4-6 is controversial but preferred to these other historical terms, which have largely been used in histopathology.

“Poorly differentiated carcinoma” is a poorly defined lesion but in most series reflects a primary thyroid carcinoma without papillary or follicular or other differentiation and lacks the severe anaplasia of anaplastic carcinoma, but the distinction in each case reflects a spectrum of dysplasia. Metastases to thyroid can produce a number of different patterns and are included in the DD of high-grade primary carcinomas.

The eight thyroid FNAB smear patterns are presented in Table 4-5 .

In terms of the Bethesda System of reporting, in which the diagnoses are based on both low-power and high-power assessment, smear patterns 1, 2, 3, and 4 broadly correlate with the “benign” category, while smear pattern 5 correlates with “suspicious of a follicular neoplasm” or “suspicious of carcinoma,” and smear pattern 6, 7, and 8 correlate with “suspicious of carcinoma” or “carcinoma.” The smear pattern is the initial assessment of the FNAB material, and the DD of each smear requires high-power assessment to decide which of the differential diagnoses of that pattern is present.

Plate 4-1

Pattern 1 Inflammatory Cells Predominate

Plate 4-1 Table

Pattern 1: Inflammatory Cells Predominate.

Entities in the Differential Diagnosis i. Suppurative with or without organism a. Bacterial b. Fungal ii. Granulomatous a. Subacute granulomatous thyroiditis (De Quervain) b. Mycobacterial thyroiditis iii. Lymphoid a. Chronic lymphocytic thyroiditis/Hashimoto thyroiditis b. Mucosa-associated lymphoid tissue (MALT) lymphoma c. Monotonous small to intermediate lymphoid cells: small cell lymphomas d. Monotonous intermediate to large lymphoid cells: intermediate to large cell lymphomas

Plate 4-2

Pattern 2 Cystic Background With Evidence of Old Hemorrhage

Plate 4-2 Table

Pattern 2: Cystic Background With Evidence of Old Hemorrhage

Entities in the Differential Diagnosis i. With colloid and follicular tissue fragments ii. With papillary tissue fragments

Plate 4-3

Pattern 3 Predominant Thin Colloid With Scant Follicular Epithelial Sheets: Colloid Nodule

Plate 4-3Table

Pattern 3: Predominant Colloid With Scant Follicular Epithelial Sheets: Colloid Nodule

Entities in the Differential Diagnosis i. Colloid nodule

Plate 4-4

Pattern 4 Follicular Tissue Fragments and Larger Tissue Fragments in Which Fibrovascular Strands Separate Follicles of Varying Size in a Background of Usually Abundant Thin Colloid

Plate 4-4 Table

Pattern 4: Follicular Tissue Fragments and Larger Tissue Fragments in Which Fibrovascular Strands Separate Follicles of Varying Size, in a Background of Usually Abundant Thin Colloid

Entities in the Differential Diagnosis i. Normal thyroid ii. Benign follicular nodule iii. Hyperplastic thyroid disease (Graves Disease)

Plate 4-5

Pattern 5 Epithelial Tissue Fragments Predominantly Microfollicles and Scant Colloid

Plate 4-5 Table

Pattern 5: Epithelial Tissue Fragments Predominantly Microfollicles and Scant Colloid

Entities in the Differential Diagnosis i. Follicular neoplasm: follicular adenoma or low grade follicular carcinoma with capsular or vascular invasion ii. Follicular variant of papillary carcinoma iii. Parathyroid tumor

Plate 4-6

Pattern 6 Epithelial Tissue Fragments With Granular Cytoplasm and Scant or No Colloid

Plate 4-6 Table

Pattern 6: Epithelial Tissue Fragments With Granular Cytoplasm and Scant or No Colloid.

Entities in the Differential Diagnosis i. Oncocytic neoplasm: Hurthle cell adenoma or carcinoma ii. Oncocytic variant of follicular carcinoma iii. Oncocytic variant of papillary carcinoma iv. Medullary carcinoma. v. Metastatic carcinoma including renal cell carcinoma

Plate 4-7

Pattern 7 Epithelial Tissue Fragments With Well-Defined Cytoplasm, Nuclear Grooves and Pseudoinclusions, and Variable Papillary Tissue Fragments and Variable Colloid

Plate 4-7 Table

Pattern 7: Epithelial Tissue Fragments With Well-Defined Cytoplasm, Nuclear Grooves and Pseudoinclusions, and Variable Papillary Tissue Fragments and Variable Colloid.

Entities in the Differential Diagnosis i. Papillary carcinoma ii. Hyalinizing trabecular tumor

Plate 4-8

Pattern 8 Predominantly Single Cells and Epithelioid or Spindle Cells

Plate 4-8 Table

Pattern 8: Predominantly Single Cells With Epithelioid or Spindle Cells

Entities in the Differential Diagnosis i. Medullary thyroid carcinoma ii. Poorly differentiated ("insular") carcinoma of thyroid iii. Anaplastic carcinoma of thyroid iv. Metastatic tumors including melanoma and lobular breast carcinoma.