Fax: (617) 573-3389
The atypical thyroid fine-needle aspiration: Past, present, and future
Version of Record online: 12 JUL 2011
Copyright © 2011 American Cancer Society
Volume 120, Issue 2, pages 73–86, 25 April 2012
How to Cite
Bongiovanni, M., Krane, J. F., Cibas, E. S. and Faquin, W. C. (2012), The atypical thyroid fine-needle aspiration: Past, present, and future. Cancer Cytopathology, 120: 73–86. doi: 10.1002/cncy.20178
- Issue online: 13 APR 2012
- Version of Record online: 12 JUL 2011
- Manuscript Accepted: 6 JUN 2011
- Manuscript Revised: 3 JUN 2011
- Manuscript Received: 9 MAY 2011
- fine-needle aspiration;
- atypia of undetermined significance/follicular lesion of undetermined significance (AUS/FLUS);
Thyroid fine-needle aspiration has developed into a key test in the evaluation of thyroid nodules. Although the interpretation of thyroid aspirates containing mild abnormalities is problematic, the introduction of the atypia of undetermined significance/follicular lesion of undetermined significance (AUS/FLUS) category in The Bethesda System for Reporting Thyroid Cytopathology (TBSRTC) has helped to delineate such cases in a systematic and clinically meaningful manner. Herein the authors review the cytomorphologic features associated with the AUS/FLUS interpretation and summarize the results of studies conducted since the implementation of TBSRTC. Cancer (Cancer Cytopathol) 2012;. © 2011 American Cancer Society.
Fine-needle aspiration (FNA) is the primary testing modality for identifying malignancy in patients with a thyroid nodule. A majority of thyroid FNA specimens, generally in the range of 60% to 70%, are classified as “benign,” and approximately 20% to 30% in total fall into the 3 categories of “suspicious for a follicular neoplasm” (SFN), “suspicious for malignancy,” and “malignant.”1, 2 The approximately 10% of cases remaining represent a small but significant subset of thyroid FNA specimens that exhibit some form of atypia of uncertain significance.3 Such atypia most often results from a variety of benign cellular changes, but in some instances it reflects an underlying malignancy that has been suboptimally sampled or has subtle diagnostic features.
Historically, the reporting of borderline atypia in thyroid aspirates was problematic. Diagnostic terms such as “atypia” and “indeterminate” had been used by different laboratories to mean different things, hindering communication and making comparison between studies difficult.4-7 For example, the term “indeterminate” had been used to mean 1 or more of the following: atypical cells, not otherwise specified (NOS); a predominantly microfollicular lesion; follicular neoplasia; and atypical cells, rule out papillary thyroid carcinoma (PTC).8-11
Definition of the Atypia of Undetermined Significance/Follicular Lesion of Undetermined Significance Category
The recently introduced Bethesda System for Reporting Thyroid Cytopathology (TBSRTC)12 named and defined a specific category for borderline cases that do not fulfill the criteria for the other defined categories. TBSRTC offers a choice of 2 terms for this category: “atypia of undetermined significance” (AUS) and “follicular lesion of undetermined significance” (FLUS). The terms AUS and FLUS are synonymous because there was no consensus reached at the National Cancer Institute Thyroid Fine-Needle Aspiration State of the Science Conference regarding a single name for all categories, and laboratories are expected to choose the one they prefer for reporting thyroid FNA interpretations (Table 1). According to TBSRTC, the AUS/FLUS category is reserved for aspirates that contain follicular, lymphoid, or other cell types with architectural and/or nuclear atypia that is more pronounced than that observed in benign/reactive lesions yet not sufficient to be classified as SFN, suspicious for malignancy, or malignant.3, 12
|Diagnostic Category||Risk of Malignancy (%)||Usual Management|
|I. Nondiagnostic or unsatisfactory||1-4||Repeat FNA with ultrasound guidance|
|II. Benign||0-3||Clinical follow-up|
|III. Atypia of undetermined significance or follicular lesion of undetermined significance||5-15||Repeat FNA|
|IV. Follicular neoplasm or suspicious for a follicular neoplasm||15-30||Surgical lobectomy|
|V. Suspicious for malignancy||60-75||Near-total thyroidectomy or surgical lobectomy|
|VI. Malignant||97-99||Near-total thyroidectomy|
By definition, AUS/FLUS is a heterogeneous category that includes a variety of abnormal architectural, cellular, or nuclear features and, as such, implementation of this category among cytopathologists, at least initially, is expected to be variable. As with other diagnostic categories of uncertainty (such as “atypical squamous cells of undetermined significance” [ASCUS] for cervical cytology), AUS/FLUS has the potential to be overused. At the time of its implementation, the authors of the TBSRTC recommended that the AUS/FLUS category should not exceed 7% of thyroid FNA diagnoses. Based on available studies, the associated risk of malignancy for this category was anticipated to be in the range of 5% to 15%,13-15 intermediate between that of the benign and suspicious categories (Table 1) (Table 2).7, 15-22 In most instances, TBSRTC indicated that the initial follow-up of thyroid aspirates diagnosed as AUS/FLUS should be a repeat FNA (Table 1). In the majority of cases, a repeat FNA resolves the atypia into a more definitive diagnostic category.
|Reference||AUS/FLUS Rate, %||Frequency of rFNA After AUS/FLUS, %||Persistent AUS/FLUS Cases in rFNA, %||Most Common Diagnosis in rFNA (%)||AUS/FLUS Cases With Histologic Follow-Up, %||Malignancy Rate for AUS/FLUS (Analysis Restricted to Cases With Histologic Follow-Up), %|
|Nayar & Ivanovic 200917||18 (total)||21 (total)||27 (total)||Benign (58) (total)||46.5 total||6 (total)|
|Theoharis 200921||3||19c||6||Benign (65)||30.3c||48c|
|Ohori 201016||20.5||49d||67c||Benign (84)c||24||17.1e|
|Faquin & Baloch 201019||9 (MGH)||40||23||Benign (62)||54f||19f|
|Jo 201022||3.4||8||0||Nondiagnostic (62)||52||17|
|VanderLaan 201118||10.9||56.1||27.9||Benign (48.4)||38.9||27.5g|
TBSRTC outlines a variety of scenarios for which the AUS/FLUS category is appropriate.3 These scenarios are not comprehensive, but they do include the most common situations encountered in clinical practice that result in a diagnosis of AUS/FLUS. Definitions and examples are reviewed below. It is worth noting that, in many instances, a predisposing condition for an AUS/FLUS interpretation is a compromised specimen (eg, sparse cellularity, obscuring blood, or poor fixation).3
An abnormal arrangement of follicular cells with each other, so-called “architectural atypia,” can raise concern for a neoplasm under several different circumstances without being sufficiently convincing to warrant the interpretation of SFN. A subset of microfollicles is normal in a benign thyroid FNA specimen; many benign thyroid nodules demonstrate a mixture of macrofollicles and microfollicles, but usually with a predominantly macrofollicular cytoarchitectural pattern. However, a subset of thyroid FNA specimens are paucicellular and contain only a few well-formed microfollicles, trabeculae, or crowded groups (Figs. 1A-1C). Some of these cases may not even meet minimal adequacy criteria of TBSRTC (at least 6 groups of follicular cells with 10 cells per group), yet to interpret the aspirate as “nondiagnostic” would ignore the architectural atypia that is present. Another pattern of concerning architectural atypia occurs when only a subset of smears (such as a single slide from 1 pass) exhibits a predominance of microfollicles whereas other smears from the same case demonstrate a benign macrofollicular pattern. Still another instance of architectural atypia is subtle but persistent crowding of follicular cells throughout a paucicellular aspirate (Fig. 1D).
Several patterns of nuclear atypia raise the specter of PTC but are quantitatively and/or qualitatively insufficient for an interpretation of “suspicious for malignancy.” These include a hypocellular aspirate (sometimes in the setting of extensive cystic degeneration) with only rare follicular cells exhibiting nuclear atypia suggestive of PTC (Fig. 2A). Alternatively, similar rare cells may be present in a cellular background that has an otherwise benign appearance, obscuring the significance of the rare atypical cells (Fig. 2B). In some instances, more diffuse but mild nuclear changes can be appreciated, with nuclear enlargement, crowding, and pallor, but without other supporting features (eg, intranuclear pseudoinclusions, grooves, and nuclear contour irregularities) of PTC (Fig. 2C). Focal and/or mild changes that suggest PTC are occasionally encountered in settings known to mimic the cytologic atypia of PTC, therefore raising doubt about the significance of the findings. Such confounding situations include patients with chronic lymphocytic thyroiditis (Hashimoto thyroiditis [HT]) (Fig. 2D) and those with a history of external beam radiation, radioiodine therapy (Fig. 2E), or thyroid suppressive agents, most notably for Graves disease.
Oncocytic patterns of uncertain significance
Some paucicellular thyroid aspirates are comprised of a virtually exclusive population of oncocytes (grouped or isolated) while lacking lymphocytes or significant numbers of benign, nononcocytic follicular cells. The paucicellularity mitigates against the interpretation of “SFN, Hurthle cell type” (SFNHCT), but the exclusive or virtually exclusive population of Hurthle cells precludes an interpretation of “benign” (Figs. 3A and 3B). Aspirates with this pattern are better placed in the AUS/FLUS category, and patients may benefit from a repeat FNA performed within 3 to 6 months.23, 24
Another pattern is the cellular aspirate comprised exclusively or nearly exclusively of oncocytes in a patient with chronic lymphocytic thyroiditis (HT) (Fig. 3C) or multinodular hyperplasia (MNH). In both clinical situations, it is not unusual to find hyperplastic nodules comprised exclusively of Hurthle cells. If the aspirate is comprised exclusively or nearly exclusively of oncocytes and clinical data or ultrasound features suggest HT, either SFNHCT or AUS/FLUS is an appropriate interpretation. If the pathologist chooses AUS/FLUS, the clinician is thereby informed that the differential diagnosis includes an oncocytic hyperplasia. Whereas an SFNHCT interpretation would prompt a reflex lobectomy, an interpretation of AUS/FLUS allows for clinical and radiologic correlation, with the possibility of conservative management by means of periodic thyroid examinations in selected circumstances. Roh et al recently demonstrated that the risk of malignancy in a patient with HT and an FNA specimen comprised of a pure population of Hurthle cells falls halfway between the malignancy risks associated with benign and SFNHCT interpretations.25 Classification of a cellular Hurthle cell aspirate in the setting of MNH is similarly flexible. TBSRTC provides for either using the AUS/FLUS category or the SFNHCT category in this situation, based on the judgment of the cytopathologist. Clinical and radiologic correlation rather than repeat FNA are more likely to determine whether surgical intervention is appropriate in such cases.
Atypia secondary to preparation artifact
Preparation artifact alone is not sufficient to warrant the interpretation of AUS/FLUS. Nevertheless, some thyroid FNA specimens contain atypical cells that are hard to observe and assess confidently because of obscuring blood, poor clotting artifact, or poor fixation or staining (Figs. 4A-4C). Excessive blood and microclots can obscure the architectural pattern of follicular cells, in some instances making the groups appear falsely crowded or microfollicular.11
When improperly fixed or stained slides are used, cells may appear larger than usual, especially at the periphery of the smear. Nuclei can appear enlarged, and the chromatin can resemble the “salt-and-pepper” pattern observed in medullary thyroid carcinoma. True atypical cells (with grooves and nuclear enlargement) can be admixed with such artifacts and lead to a false-positive interpretation. Similarly, if the stain is very pale, the chromatin pattern can raise the possibility of PTC.
Atypical cyst-lining cells
Atypical cyst-lining cells are a potential pitfall encountered with cystic thyroid nodules.26 Benign cyst-lining cells exhibit a range of appearances from elongate, reactive-appearing cells to markedly enlarged cells with nuclear features that mimic PTC. Benign cyst-lining cells are typically polygonal or fusiform, with abundant dense cytoplasm; well-defined cellular borders; occasionally enlarged, grooved nuclei; and small, distinct nucleoli (Fig. 5A). In the context of a cystic aspirate with admixed macrofollicles and colloid, the benign cyst-lining cells are easily recognized as benign. However, in isolation, it can be difficult to exclude PTC. In the latter instance, the aspirate is better classified as AUS/FLUS to reflect the uncertainty regarding the findings (Figs. 5B-5D).
Other nuclear atypias
In some situations, a minor population of follicular cells contain nuclei that are markedly enlarged with prominent nucleoli. Such focal atypical features can be observed in specimens from patients with a history of treatment with radioactive iodine (Figs. 6A and 6B), carbimazole, or other pharmaceutical agents, and can also be noted secondary to reparative changes associated with hemorrhage or cystic degeneration.3, 27, 28 Knowledge of the clinical history can help to avoid overdiagnosis, but the atypia may be sufficiently marked that it warrants placing the diagnosis into the AUS/FLUS category.
In this scenario, the thyroid aspirate is comprised of a population of atypical lymphocytes (Fig. 7A) that nevertheless lack sufficient atypia to be interpreted as suspicious for malignancy. The lymphocytes may be small to intermediate in size, with mild nuclear atypia such as occasional cleaves or small nucleoli, as can be observed in some low-grade lymphoproliferative disorders (eg, extranodal marginal zone lymphoma of mucosa-associated lymphoid tissue [MALT] type). Ideally, material for flow cytometry is available to help clarify the nature of the atypical lymphocytes, but for cases limited to microscopic evaluation alone, a diagnosis of AUS/FLUS is appropriate, with a recommendation for additional sampling for immunophenotyping. This scenario is in contrast to most cases of diffuse large B-cell lymphoma of the thyroid, which can readily be diagnosed as being suspicious for malignancy or malignant by FNA (Fig. 7B).
Scenarios “not otherwise specified”
The scenarios described above account for nearly all AUS/FLUS cases, but other, less common situations are encountered that are of uncertain diagnostic significance and warrant the AUS/FLUS designation. For example, the cytologist may encounter a population of atypical cells of uncertain origin (?follicular vs parafollicular,?mesenchymal vs epithelial,?primary tumor vs metastatic malignancy) (Figs. 8A and 8B). In the absence of available ancillary material to resolve this diagnostic uncertainty, an AUS/FLUS interpretation is appropriate.
One of the seminal accomplishments of TBSRTC is the introduction of a uniform terminology to facilitate the study of thyroid FNA specimens through improved communication and the reliable sharing of data among the cytology community. Because only a short interval of time has passed since the implementation of TBSRTC, published experience with AUS/FLUS is inevitably limited, and one must be wary of attempts to retrofit laboratory experiences to TBSRTC. With this caveat in mind, published experience with the AUS/FLUS category is reviewed below.
TBSRTC recommends that the rate of AUS/FLUS interpretations within a given laboratory not exceed 7%. It is important to note that this figure was intended as an initial guideline to encourage judicious use of the category and was not based on extensive clinical experience. The percentage of cases diagnosed as AUS/FLUS has been shown to vary widely. Although most published series demonstrate rates within the range of 7% to 12%, there is broad variation, with rates as low as 3% and as high as 20.5% (Table 2). Not surprisingly, those laboratories with higher AUS/FLUS rates typically report lower rates of malignancy on follow-up.16, 17 These higher rates appear to be attributable in part to a relative over-representation of cases with poor cell preservation that might be better classified as benign or nondiagnostic.16 Conversely, it has been suggested that extremely low rates of AUS are likely achieved at the expense of decreased sensitivity.29 Overall, although the 7% target for AUS/FLUS may remain a reasonable goal, current experience suggests that slightly higher rates, in the range of 7% to 12%, are likely to be more typical, at least in the early years of implementation of TBSRTC.
In view of the various scenarios included in the AUS/FLUS diagnostic category, it is not surprising that there are significant interobserver and interinstitutional variabilities in making an AUS/FLUS diagnosis. Layfield et al, Ohori et al, Muddegowda et al, and Shi et al have reported low reproducibility for AUS/FLUS, with marked interobserver and intraobserver variations.15, 16, 30, 31 The level of experience of the clinician/pathologist performing the thyroid FNA, the methods used for specimen preparation and staining, the number of FNA passes performed, and the availability of onsite rapid interpretation may all be contributing factors to this variation. The publication of reference images and text for TBSRTC, both in print and on the Internet (available at: www.papsociety.org), will inevitably promote greater uniformity. Other methods for improving performance have also been proposed. These include educational sessions using standardized and clear diagnostic criteria,32 the use of representative case materials,33 the use of consensus review with cytological-histological correlation,34 and providing periodic individual feedback through quality assurance metrics.35 These methods may reduce the rate of AUS/FLUS interpretations and the associated variability without compromising diagnostic performance.
Clinical significance of the AUS/FLUS diagnosis
The malignancy rate for the AUS/FLUS category was estimated to be between 5% and 15%.12 Here, too, there has been wide variation in reported experience (Table 2). In studies that include cytologic follow-up, the rate of malignancy has usually been demonstrated to be within this low range, although the rate was reported to be 27.5% in 1 large study.18 In the selected subset of patients undergoing surgery, the rate of malignancy is reported to vary between 6% and 48% (Table 2). Nevertheless, in the majority of laboratories, the average rate of malignancy for AUS/FLUS is intermediate between that of the benign category (0%-3%) and that of the SFN category (15%-30%). The most common malignant diagnosis made at the time of surgery in cases initially diagnosed as AUS/FLUS is PTC, usually of the follicular variant (PTC-FV).15, 19, 36-38 PTC-FV is a challenging diagnosis, often exhibiting subtle nuclear features (both cytologically and histologically), with less-than-ideal reproducibility.39 This likely accounts for at least some of the observed variability in malignancy rates.
Initial management of AUS/FLUS cases
Given the anticipated low rate of malignancy associated with an initial AUS/FLUS diagnosis, TBSRTC recommends that most patients undergo a repeat thyroid FNA within 3 to 6 months in an attempt to better define the nature of the atypia.3, 19, 20, 32 An exception is the pure oncocytic pattern of AUS as outlined and discussed earlier. This approach to managing patients with an AUS/FLUS diagnosis is considered safe and cost-effective.40, 41 Correlation with clinical and radiologic features is also important, because patient management, although heavily influenced by the FNA interpretation, can and should be impacted by clinical and sonographic findings, and even the patient's own desires for treatment. In approximately 20% to 28% of AUS/FLUS cases, a repeat thyroid FNA will again be interpreted as AUS/FLUS (Table 2).3, 18, 42 The risk of malignancy in this subset of patients has been reported to be higher by some19 but not others,17 and this approach has been called into question.18 Furthermore, one of the assumptions underlying this approach is that a benign aspirate after a diagnosis of AUS/FLUS carries the same low risk of malignancy as an isolated “benign” aspirate, but the validity of this assumption has been challenged by some authors.18, 43
Qualifiers of AUS
As outlined above, it is clear that the AUS/FLUS category is a heterogeneous one. As such, it is to be expected that different patterns of AUS/FLUS may carry different risks of malignancy. Although qualifying AUS/FLUS is not explicitly required by TBSRTC, some laboratories do so, in one way or another, and some authors have retrospectively reviewed cases in an attempt to correlate patterns with malignancy rates. Data have suggested that the presence of focal cytologic features of PTC confers a higher risk of malignancy than other AUS/FLUS patterns,20, 44 whereas architectural atypia alone is approximately half as likely to be malignant as other patterns and is more likely to be a follicular adenoma.36 Among the challenges of an approach that uses qualifiers is the ability to reliably distinguish these patterns from one another as well as other TBSRTC categories (eg, AUS/FLUS with cytologic atypia or focal features of PTC from the “suspicious for malignancy” category). Currently, further study of AUS/FLUS qualifiers for the purposes of refining criteria for this category and further stratifying the risk associated with different patterns has value much in the way that studying qualifiers of ASCUS helped to refine successive implementations of The Bethesda System for Reporting Cervical Cytology.45
Why use the AUS/FLUS category?
The finding that the risk of malignancy associated with the AUS/FLUS category is intermediate between the benign and SFN categories supports the use of AUS/FLUS as a separate and distinct category within TBSRTC (Table 1). Some authors have proposed that the risk of malignancy associated with AUS/FLUS sufficiently approximates the SFN category so that the use of a separate AUS/FLUS category is unnecessary.37, 46 However, the heterogeneity of AUS/FLUS may still result in greater refinement of this category, something that would be lost by grouping it with SFN. Furthermore, the malignancies most associated with the categories are different (PTC for AUS/FLUS, follicular carcinoma for SFN), which may influence clinical decision-making regarding surgery. In addition, Shi et al and Yang et al have shown that if AUS cases are forced into either lower or higher diagnostic risk categories, the latter results in diminished sensitivity for detecting thyroid neoplasms, with increased false-negative and false-positive rates, making thyroid FNA a less effective screening test.31, 47
Although to our knowledge immunohistochemistry has only a very limited role in thyroid cytology,48-50 molecular testing using markers such as BRAF, RET/PTC, RAS, and paired box gene (PAX)/peroxisome proliferator-activated receptor-γ (PPAR-γ) shows more promise for wider applicability.51-54 A BRAF mutation or RET/PTC gene rearrangement has very high specificity for PTC. When these molecular markers are positive in the AUS/FLUS category, the probability of a cancer outcome is very high.16, 55 However, a negative molecular result using the panel above is much less useful, in large part because many tumors, particularly PTC-FV, are negative for all of them or positive only for RAS mutations, which have much lower specificity and are noted in benign and malignant thyroid neoplasms.56, 57 A major obstacle to the application of a limited panel of molecular markers to thyroid FNA is the finding that currently available markers tend to work well with lesions that are readily diagnosed by cytomorphology; they are not as useful for the difficult thyroid FNA lesions diagnosed as AUS/FLUS, SFN, or suspicious for malignancy (PTC-FV). More recently, microarray data from > 200 genes have been used to produce a “benign thyroid fingerprint” that has the potential for use in guiding the management of patients with an AUS/FLUS interpretation.58
AUS/FLUS is comprised of a heterogeneous group of FNA patterns that are not clearly benign, suspicious, or malignant. Although the reproducibility of this interpretation is (fair? poor? fair-poor?), its use is warranted, but care must be exercised that it is used judiciously, and laboratories are encouraged to monitor the frequency with which the interpretation is made. The recommended management of an initial AUS is a repeat FNA in most instances. As experience is gained with TBSRTC in general and AUS/FLUS in particular, it is expected that the criteria for AUS/FLUS as well its clinical management may be refined further. It is likely that molecular testing will one day play an important role in triaging patients with a diagnosis of AUS/FLUS.
No specific funding was disclosed.
CONFLICT OF INTEREST DISCLOSURES
Dr. Cibas is on the Steering Committee/Writing Group of Veracyte.
- 12Ali SZ, Cibas ES, eds. The Bethesda System for Reporting Thyroid Cytopathology: Definitions, Criteria and Explanatory Notes. New York: Springer; 2010.
- 14Diagnostic terminology and morphologic criteria for cytologic diagnosis of thyroid lesions: a synopsis of the National Cancer Institute Thyroid Fine-Needle Aspiration State of the Science Conference. Diagn Cytopathol. 2008; 36: 425-437., , , et al.
- 16Contribution of molecular testing to thyroid fine-needle aspiration cytology of “follicular lesion of undetermined significance/atypia of undetermined significance.” Cancer (Cancer Cytopathol). 2010; 118: 17-23., , , et al.
- 17The indeterminate thyroid fine-needle aspiration: experience from an academic center using terminology similar to that proposed in the 2007 National Cancer Institute Thyroid Fine Needle Aspiration State of the Science Conference. Cancer (Cancer Cytopathol). 2009; 117: 195-202., .
- 29Subclassification of atypical cells of undetermined significance in direct smears of fine-needle aspirations of the thyroid: distinct patterns and associated risk of malignancy [published online ahead of print March 25, 2011]. Cancer (Cancer Cytopathol). doi: 10.1002/cncy. 20154..
- 30Divide and rule: cytodiagnosis of thyroid lesions using pattern analysis: a study of 233 cases [published online ahead of print November 9, 2010]. Diagn Cytopathol., , , .
- 32Minimizing the diagnosis of “follicular lesion of undetermined significance” and identifying predictive features for neoplasia [published online ahead of print October 14, 2010]. Diagn Cytopathol., , , , .
- 33Thyroid FNA atypia of undetermined significance (AUS): variability in pathologist reporting and clinical follow-up. Mod Pathol. 2011; 24( 1 suppl): 98A., , , et al.
- 34Consensual review minimizes the diagnosis of “follicular lesion of undetermined significance” and improves reproducibility and cyto-histologic concordance. Mod Pathol. 2011; 24( 1 suppl): 94A., , , et al.
- 35Atypia of undetermined significance in thyroid fine-needle aspiration: characterizing cytopathologist practice patterns. Mod Pathol. 2011; 24( 1 suppl): 108A., , .
- 36Utility of diagnostic qualifiers for thyroid fine-needle aspirations with atypia of undetermined significance. Am J Clin Pathol. In press., , .
- 38Follicular lesions of the thyroid: a retrospective study of 1,348 fine needle aspiration biopsies [published online ahead of print October 17, 2010]. Diagn Cytopathol., , , , .
- 45The Bethesda System for Reporting Cervical Cytology. 2nd ed. New York: Springer-Verlag; 2004., .
- 53Differential diagnosis of thyroid nodules using fine-needle aspiration cytology and oncogene mutation screening: are we ready? F1000 Med Rep. 2010; 2: 62., , .
- 58Development of a novel molecular classifier to accurately identify benign thyroid nodules in patients with indeterminate FNA cytology. Presented at the 14th International Thyroid Congress; September 11-16, 2010; Paris, France., , , et al.