Fine-needle aspiration of thyroid nodules: A study of 4703 patients with histologic and clinical correlations

Authors


Abstract

BACKGROUND.

The Papanicolaou Society of Cytopathology recently proposed 6 diagnostic categories for the classification of thyroid fine-needle aspiration (FNA) cytology. Using these categories, the experience with FNA from 2 institutions was studied with emphasis on cytologic-histologic correlation, source of errors, and clinical management.

METHODS.

Patient cytology data were retrieved by a retrospective search of thyroid FNA in the institutional databases. Cytologic diagnoses were classified as unsatisfactory, benign, atypical cellular lesion (ACL), follicular neoplasm (FN), suspicious for malignancy, and positive for malignancy. Samples with a histologic discrepancy were re-evaluated, and clinical follow-up information was recorded.

RESULTS.

Of of 4703 FNA samples, 10.4% were classified as unsatisfactory, 64.6% were classified as benign, 3.2% were classified as ACL, 11.6% were classified as FN, 2.6% were classified as suspicious, and 7.6% were classified as malignant. Five hundred twelve patients had at least 1 repeat FNA, mainly for results in the unsatisfactory and ACL categories. One thousand fifty-two patients had surgical follow-up, including 14.9% of patients with unsatisfactory FNA results, 9.8% of patients with benign results, 40.6% of patients with ACL results, 63.1% of patients with FN results, 86.1% of patients with suspicious results, and 79.3% of patients with malignant results. The rates for histologically confirmed malignancy in these categories were 10.9%, 7.3%, 13.5%, 32.2%, 64.7%, and 98.6%, respectively. The cytologic-histologic diagnostic discrepancy rate was 15.3%. Sources of errors included diagnoses on inadequate specimens, sample errors, and overlapping cytologic features between hyperplastic nodules and follicular adenoma. The sensitivity and specificity of thyroid FNA for the diagnosis of malignancy were 94% and 98.5%, respectively.

CONCLUSIONS.

The current results indicated that FNA provides an accurate diagnosis of thyroid malignancy. The 6 diagnostic categories were beneficial for triaging patients for either clinical follow-up or surgical management. Cancer (Cancer Cytopathol) 2007. © 2007 American Cancer Society.

Fine-needle aspiration (FNA) of the thyroid has been accepted as a first-line screening test for patients with thyroid nodules. Its utility has increased significantly in recent years, partially because of the availability of ultrasound-guided techniques, which allow for the detection and aspiration of smaller and deep-seated nodules. Consequently, more thyroid cancers are diagnosed at an early stage.1 The main role of thyroid FNA is to triage patients for either surgery or conservative management. Patients with FNA diagnoses that suggest malignancy and/or neoplasia are managed surgically, whereas patients with FNA diagnoses that favor a benign lesion can be followed clinically. Currently, worldwide, there is no standardized terminology for the cytologic diagnosis of thyroid FNA. Diagnoses and diagnostic categories are defined differently among institutions and even between individuals in the same institution. This is especially true for use of the indeterminate category.2–6 This lack of uniform reporting may confuse clinicians and can result in inconsistent patient management. In an effort to establish standard terminology and a uniform approach for the reporting of thyroid cytology, the Papanicolaou Society of Cytopathology recently developed recommendations for the assessment of specimen adequacy, diagnostic criteria, and reporting format,7 following the 1996 guidelines for the examination of thyroid FNA specimens.8 The new recommendations included 6 diagnostic categories for thyroid FNA cytology: unsatisfactory, benign, cellular follicular lesion, follicular neoplasm (FN), suspicious for malignancy, and positive for malignancy.7 These 6 categories are similar to those that we have been using for several years. In the current report, we have summarized the experience of 2 institutions in using these 6 categories for diagnosing thyroid FNA. Emphasis is placed on cytologic-histologic correlation, identification of the source of errors, and clinical management of patients. The objective of this study was to provide useful information on cytologic diagnosis and clinical management of patients with thyroid nodules.

MATERIALS AND METHODS

Cases were retrieved by a computerized, retrospective search of all thyroid FNA specimens filed in the Department of Pathology, Long Island Jewish Medical Center (New York, NY) between January 1992 and December 2003 and the Department of Pathology, The University of Texas Medical Branch (Galveston, Tex) between January 1993 and May 2005. For most patients, both rapid Romanowsky (Diff-Quick)-stained and Papanicolaou-stained smears were available for review. The cytologic diagnoses were classified into 6 categories with the following criteria.

1. Unsatisfactory

The specimens that were categorized as unsatisfactory consisted predominantly of blood, lacked both colloid and follicular cells, and/or had <6 groups of well-preserved cells on each of at least 2 slides. Hypocellular aspirates from cysts that contained fluid with either abundant colloid or macrophages were not considered unsatisfactory if they correlated with the clinical impression. In addition, a specimen that contained obvious abnormal cells was not deemed unsatisfactory regardless of the quantity of epithelial cells observed.

2. Benign/Negative for Neoplasia

Specimens that were categorized as benign/negative for neoplasia included 1) specimens with abundant colloid, variable numbers of unremarkable follicular cells in flat sheets and macrofollicles; fluid containing various numbers of macrophages, colloid, and small, uniform follicular cells with coarse chromatin; and 2) specimens with lymphocytes, Hurthle cells, or other inflammatory cells. The differential diagnosis for this group mainly included thyroid cyst, colloid nodule, nodular hyperplasia, and Hashimoto thyroiditis.

3. Atypical Cellular Lesion

Specimens composed of variable amount of colloid and a large number of follicular cells in flat sheets with macrofollicles and microfollicles were classified as atypical cellular lesions (ACL). In these specimens, the pathologist favored a benign, hyperplastic nodule; however, FN could not be excluded completely. Also included in this category were specimens with focal atypical cellular features, such as enlarged nuclei with irregular nuclear membranes and pale chromatin, in which the possibility of a malignant process, especially papillary carcinoma of the thyroid (PCT), could not be ruled out completely.

4. FN/Indeterminate

The FN/indeterminate category applied to cellular specimens with scant or absent colloid and follicular cells or Hurthle cells that formed syncytial or thick, 3-dimensional clusters; loosely cohesive microfollicles; and isolated, intact, individual cells. FN nuclei generally were overlapping, enlarged, round, and lacked typical nuclear features of PCT. The differential diagnosis for this category was mainly follicular/Hurthle cell adenoma (FA/HA), follicular/Hurthle cell carcinoma (FC/HC), and the follicular variant of papillary carcinoma (FVPCT).

5. Suspicious for Malignancy

Most specimens in this category had some cytologic features suggestive of papillary carcinoma, such as enlarged, atypical nuclei with nuclear grooves; intranuclear pseudoinclusions; and fine, granular chromatin. However, these features were not quantitatively or qualitatively sufficient to make a definitive diagnosis of malignancy. This category was used mainly for specimens that were suspicious for PCT. A few specimens that were interpreted near the beginning of the study period were reported as suspicious for follicular carcinoma because of the categorization at 1 institution of Hurthle cell neoplasms (HN) and FN with marked atypia as suspicious for malignancy during the early part of the 1990s.

6. Positive for Malignancy

Specimens that had cytologic features consistent with malignant neoplasms, including PCT, medullary carcinoma, anaplastic carcinoma, lymphoma, and metastatic cancers, were classified as positive for malignancy.

The specimens with discrepant cytologic and histologic diagnoses were reviewed and re-evaluated to identify the source of errors. The sensitivity and specificity of cytologic diagnosis were evaluated with 2 different strategies, both using the histologic diagnosis as the gold standard: 1) testing the ability of thyroid FNA to identify malignancies, in which specimens that had a cytologic diagnosis of negative and positive for malignancy were included in the calculation; 2) testing the ability of thyroid FNA to identify neoplasms, in which specimens in the benign category were considered negative; and the categories FN, suspicious for malignancy, and malignant were considered positive. The z test was used for statistical analysis.

RESULTS

In total, 4703 thyroid FNA specimens from 3337 women and girls and 612 specimens from men and boys (women/girls:men/boys, 5.4:1) were reviewed. The age of patients ranged from 2 months to 99 years (mean ± standard deviation, 49.6 ± 15.6 years). Data from the 2 study institutions were similar; therefore, they were analyzed jointly (Table 1). The distributions of initial cytologic diagnoses in the 6 diagnostic categories were as follows: 488 unsatisfactory (10.4%), 3036 benign (64.6%), 152 ACL (3.2%), 544 FN (11.6%), 124 suspicious for malignancy (2.6%), and 359 malignant (7.6%) (Table 1). The cytologic diagnoses, classified by sex and age group, are shown in Table 2 and Table 3. Five hundred twelve patients (13%) had at least 1 repeat cytology follow-up, and 1052 patients (26.6%) underwent surgical resection, which included either partial or total thyroidectomy (Table 4). The cytologic-histologic correlations are shown in Table 5. The sensitivity and specificity of thyroid FNA were 94% and 98.5%, respectively, for the diagnosis of malignancy and 89.3% and 74%, respectively, for the diagnosis of neoplasm.

Table 1. Cytologic Diagnosis by Institution
InstituteTotal No. of casesCytology category: No. of patients (%)
UnsatisfactoryBenignACLFNSuspicious for malignancyPositive for malignancy
  1. ACL indicates atypical cellular lesion; FN, follicular neoplasm; LIJ, Long Island Jewish Medical Center; UTMB, The University of Texas Medical Branch.

LIJ3338360 (10.8)2205 (66.1)98 (2.9)392 (11.8)46 (1.4)237 (7.1)
UTMB1365128 (9.4)831 (60.9)52 (3.8)152 (11.1)80 (5.8)122 (8.9)
Total4703488 (10.4)3036 (64.6)152 (3.2)544 (11.6)124 (2.6)359 (7.6)
Table 2. Cytologic Diagnosis by Sex
SexTotal No. of casesCytology category: No. of patients (%)
UnsatisfactoryBenignACLFNSuspicious for malignancyPositive for malignancy
  1. ACL indicates atypical cellular lesion; FN, follicular neoplasm.

Women and girls3997386 (9.7)2618 (65.4)139 (3.5)456 (11.4)96 (2.4)278 (7)
Men and boys70678 (11)417 (59.1)13 (1.8)89 (12.6)28 (4)81 (11.5)
Table 3. Cytologic Diagnosis by Age
Age, yTotal No. of casesCytology category: No. of patients (%)
UnsatisfactoryBenignACLFNSuspicious for malignancyPositive for malignancy
  1. ACL indicates atypical cellular lesion; FN, follicular neoplasm.

0–3052049 (9.4)314 (60.3)12 (2.3)60 (11.5)20 (3.8)65 (12.5)
31–603022303 (10)1988 (65.8)103 (3.4)333 (11)80 (2.6)215 (7.1)
>601161136 (11.7)734 (63.2)95 (8.1)93 (8)24 (2.1)79 (6.8)
All ages4703488 (10.4)3036 (64.6)152 (3.2)544 (11.6)124 (2.6)359 (7.6)
Table 4. Patients With Cytologic and/or Surgical Follow-up
Cytologic diagnosisNo. of patients (%)
TotalCytology follow-upSignificant change in diagnosis*Surgical follow-up
  • ACL indicates atypical cellular lesion; FN, follicular neoplasm.

  • *

    Change from unsatisfactory to any valid cytologic diagnosis; changes from benign to follicular neoplasm or worse and vice versa.

Unsatisfactory309141 (45.6)11746 (14.9)
Benign2526296 (11.7)27247 (9.8)
ACL12826 (20.3)652 (40.6)
FN51645 (8.7)18326 (63.1)
Suspicious for malignancy1224 (3.2)2105 (86.1)
Positive for malignancy3480 (0)0276 (79.3)
Table 5. Cytologic-histologic Correlation
Cytologic diagnosisNo. of patientsHistologic diagnosis: No. of patients
Nodular goiterHashimoto thyroiditisFAHAFCHCPCTOther lesion
  • FA indicates follicular adenoma; HA, Hurthle cell adenoma; FC, follicular carcinoma; HC, Hurthle cell carcinoma; PCT, papillary thyroid carcinoma; ACL, atypical cellular lesion; FN, follicular neoplasm.

  • *

    Two anaplastic carcinomas and 1 lymphangioma.

  • Three lymphomas and 1 parathyroid adenoma.

  • One medullary carcinoma, 1 parathyroid carcinoma, 2 metastatic renal cell carcinomas, and 1 squamous cell carcinoma.

  • §

    Parathyroid.

  • Eight medullary carcinomas, 2 anaplastic carcinomas, and 1 lymphoma.

Unsatisfactory462111530123*
Benign2471602341420134
ACL521914720280
FN3265311109481613715
Suspicious for malignancy1051549850631§
Positive for malignancy27610303125711

Of 309 patients who had unsatisfactory specimens, 141 patients (45.6%) were followed by repeat FNA; and, of those 141 patients, a diagnostic aspirate was obtained in 117 patients (83%), resulting in the following diagnoses: 79 benign (56%), 7 ACL (5%), 12 FN (8.5%), 7 suspicious for malignancy (5%), and 12 malignant (8.5%). Repeat FNA was unsatisfactory again in 24 patients (17%). Forty-six patients (14.9%) underwent surgical resection based on clinical impression and symptoms. Of these, 23 patients (50%) had benign nonneoplastic nodules, 18 patients (39.1%) were diagnosed with FA/HA, 1 patient (2.1%) was diagnosed with HC, 2 patients (4.3%) were diagnosed with PCT, and 2 patients (4.3%) were diagnosed with anaplastic carcinoma (Table 5). There were no significant differences between the sex and age groups in this category.

Thyroid nodules from 2526 patients were diagnosed cytologically as benign, and most were nodular goiters and lymphocytic (Hashimoto) thyroiditis. There were sex and age differences in this category, with lower incidence observed among men and younger age groups (Tables 2, 3). Two hundred ninety-six patients (11.7%) had at least 1 repeat FNA. Subsequent cytologic diagnoses included 16 unsatisfactory (5.4%), 243 benign (82.1%), 10 ACL (3.4%), 20 FN (6.8%), 2 suspicious for malignancy (0.7%), and 5 malignant (1.7%). Twenty-seven patients (9.1%) patients had a second FNA diagnosis that differed significantly different from the initial interpretation. This included changes from benign to FN, suspicious, or malignant. It is noteworthy that all 5 patients who had a diagnosis that changed from benign to malignant had a final diagnosis of histologically proven PCT. Two hundred forty-seven patients (9.8%) had surgical follow-up, and their final histologic diagnoses included 184 nodular goiters (74.8%), 45 FA/HA (18.3%), 2 FC (0.8%), 13 PCT (5.3%), and 3 lymphomas (1.2%) (Table 5). The discrepancy between cytologic and histologic diagnoses in this category was 25.5%, and these differences occurred mainly between hyperplastic nodule and FA.

There were 128 patients who had an initial cytologic diagnoses of ACL. Among them, 26 patients (20.3%) had repeat cytology. Subsequent cytologic diagnosis included 19 benign (73.1%), 1 ACL (3.8%), and 6 FN (23.1%). A significant change in subsequent diagnosis was noted in 6 patients (23.1%) (Table 4). Fifty-two patients (40.6%) underwent thyroidectomy, which revealed the following diagnoses: 19 nodular goiters (36.5%), Hashimoto thyroiditis in 14 patients (26.9%), 9 FA/HA (17.3%), 2 HC (3.8%), and 8 PCT (15.4%) (Table 5).

Thyroid nodules from 516 patients were diagnosed as FN/HN. Forty-five of those patients (8.7%) underwent repeat cytology, which resulted in significant changes in diagnosis in 18 patients (40%), mainly to the benign category. Three hundred twenty-six patients (63.1%) in this diagnostic category underwent thyroidectomy, which showed 53 nodular goiters (16.3%), 11 patients with Hashimoto thyroiditis (3.4%), 157 (48.2%) FA/HA, 29 FC/HC (8.9%), 71 PCT (21.8%; mostly the follicular variant), and 5 other malignancies (1.5%), including parathyroid carcinoma and metastatic carcinomas (Table 5). The discrepancy rate between cytologic and histologic diagnosis was 19.6%, and discrepancies mainly were noted in difficulties distinguishing between FA and hyperplastic nodules. There were no differences with regard to sex and age in this indeterminate category.

In total, 122 patients had thyroid nodules diagnosed as suspicious for malignancy, including 93 that were suspicious for PCT, 23 that were suspicious for FC/HC, 4 that were suspicious for lymphoma, and 1 each that was suggestive of medullary carcinoma and for anaplastic carcinoma. The diagnoses of suspicious for FC/HC were made in the early 1990s, as discussed above (see Materials and Methods). The incidence in this category was higher among men and boys patients and among younger patients (Tables 2, 3). One hundred five patients (86.1%) underwent surgical resection, which showed 15 nodular goiters (14.3%) nodular goiters, 4 patients with Hashimoto thyroiditis (3.8%), 17 FA/HA (16.2%), 5 FC/HC (4.8%), 63 PCT (60%), and 1 parathyroid gland (Table 5). Eighty-four patients who had diagnoses of suspicious for PCT were followed with surgery, including 63 patients (75%) who had PCT confirmed histologically. In 21 patients who had diagnoses of suspicious for FC, only 5 patients (23.8%) had FC confirmed histologically (Table 6).

Table 6. Cytologic-histologic Correlation in Suspicious Categories
Cytologic diagnosisNo. of patientsHistologic diagnosis: No. of patients
Nodular goiterHashimoto thyroiditisFAHAFCHCPCTOther lesion
  • FA indicates follicular adenoma; HA, Hurthle cell adenoma; FC, follicular carcinoma; HC, Hurthle cell carcinoma; PCT, papillary thyroid carcinoma.

  • *

    Parathyroid.

Suspicious for FC2131574001*
Suspicious for PCT841234110630

Thyroid nodules from 348 patients were categorized as malignant and included 3 anaplastic carcinomas (0.9%), 3 lymphomas (0.9%), 5 FC/HC (1.5%), 16 medullary carcinomas (4.6%), 302 PCT (86.8%), and 19 metastatic carcinomas (5.5%). There was a significantly greater percentage of malignancies among men and boys (men/boys vs women/girls, 11.5% vs 7%) and among younger patients (ages <30 years vs ≥30 years, 13.9% vs 7.8%) (Tables 2, 3). There was a significant increase the incidence of metastatic cancers in the older age group: 17 of 19 metastatic cancers occurred in patients aged >65 years. No patient was followed by repeat cytology. Two hundred seventy-six patients (79.3%) underwent thyroidectomy in the study institutions. The histologic diagnoses in these patients included 1 nodular goiter (0.4%), 3 FA (1.1%), 4 FC/HC (1.4%), 257 PCT (93.1%), 8 medullary carcinomas (2.8%), 2 anaplastic carcinomas (0.7%), and 1 lymphoma (0.4%).

DISCUSSION

The results of this large thyroid FNA study confirmed and reinforced most of the common knowledge regarding the usefulness of this diagnostic tool.9–15 By using 6 categories, FNA performed on 4703 thyroid nodules showed 10.4% unsatisfactory specimens, 64.6% benign thyroid nodules, 3.2% ACL, 11.6% FN, 2.6% suspicious for malignancy, and 7.6% positive for malignancy. Thyroid nodules are more common in women. There is a higher malignancy rage among men and boys and among younger patients. HC and anaplastic carcinoma occur mainly in the older age group. Younger patients have a higher incidence of PTC. Ninety percent of metastatic cancers of the thyroid occur in patients aged >65 years.

Most of the reported sensitivity and specificity of thyroid FNA ranges between 80% and 100%. This large variation is attributable mainly to the manner in which data are analyzed. Most of studies exclude the indeterminate categories (atypical, FN, and suspicious for malignancy) from the computation. According to the guidelines of Papanicolaou Society of Cytopathology and the American Association of Clinical Endocrinologists,7, 16 thyroid nodules within the indeterminate cytologic category should be managed by surgical removal, analogous to the recommendation for management in the malignant category. Several authors have suggested that,17 for clinical management purposes, the specimens in indeterminate category should be considered positive for neoplasia for the calculation for sensitivity and specificity. In our study, there were significant differences between the sensitivity and specificity for malignancy (94% and 98.5%, respectively) and for neoplasm (89.3% and 74%, respectively).

The current study demonstrated that the overall discrepancy rate between cytologic and histologic diagnosis was 15.3%. In reviewing these specimens, the single most frequent cause of false-negative cytologic diagnoses was related to sample adequacy. Of 25 patients who had a false-negative diagnosis of PCT, most diagnoses were made on inadequate or borderline cellular samples. The diagnosis of a benign thyroid lesion should be made only on adequate samples, whereas the presence of atypical cells or cellular patterns always should be addressed regardless of cellularity. Indeed, the evaluation of sample adequacy is affected by many factors in addition to cellularity.18 Clinical information, especially physical examination and imaging studies, is important in determining sample adequacy. For an ultrasonographcally proven, simple cystic nodule, follicular cells may not be required for the evaluation of adequacy. Conversely, a firm, solid nodule does require the presence of certain numbers of epithelial cells for rendering a cytologic interpretation. It is noteworthy, although not surprising, that when the aspiration is taken and the interpretation is made by the same cytopathologist, the chances of obtaining a diagnosis are increased. In addition, it has been proven that on-site adequacy evaluation can reduce inadequate samples significantly.19, 20 The issue regarding the number of follicular cells required for adequacy has been debated over the years; and, to date, no consensus has been reached.2 Until a standardized, practical guideline is established, each individual cytopathologist will have to use criteria with which he or she is comfortable.

Other sources of error that were identified in the current study included, in the order of frequency, sampling issues, cystic PCT, and misinterpretation. Regarding sampling errors, it is well known that malignancy can coexist with benign thyroid nodules. Thorough physical examination, multiple passes to cover all areas of interest, and ultrasound-guided aspiration may reduce sampling errors.19, 21 Errors caused by misinterpretation of cytologic findings in adequate samples are uncommon. Two major sources of such errors are cystic PCT and FN. There were 2 patients who had false-negative diagnoses of PCT in specimens in which the cells lacked the classic nuclear features but had the architectural pattern commonly observed in PCT. These specimens had cytologic features similar to those described by the College of American Pathologists Nongynecologic Cytology Program.22 According to our experience, the presence of clusters of epithelial cells with dense, squamous-like cytoplasm, enlarged nuclei containing multiple vacuoles, a necrotic background, and papillary structures should suggest cystic papillary carcinoma (Fig. 1). This is especially true in patients who have lymph node metastasis. Because of their cytomorphologic similarities, distinguishing between hyperplastic (adenomatous) nodule and FN sometimes is difficult and has been the subject of many studies.23–28 Cytologically, it is almost unfeasible to differentiate hyperplastic nodule from so-called macrofollicular adenoma or simple adenoma. We do not consider this a discrepancy, because it probably would not harm patients. There are cytologic features that may help in distinguishing a hyperplastic nodule versus FN. In a hyperplastic nodule, follicular cells may form large, single-layer sheets that fold over on themselves. Within these sheets of follicular cells, there are well-formed macrofollicles that often demonstrate perifollicular fibrosis or are surrounded by basement membrane-like material. Follicular cells within follicles are cohesive and often are “held” together by basement membrane and/or matrix material (Fig. 2). Conversely, FN often has syncytial clusters of follicular cells, 3-dimensional groups, and loosely cohesive follicles (especially microfollicles) and has isolated, intact follicular cells. The follicles generally have a clean background, ie, there is no obvious basement membrane or matrix-like material associated with them. The nuclei of follicular cells in FN often are enlarged (Fig. 3). Although not intended, it is unavoidable that FNA diagnoses of FN will include some specimens that prove to be FVPCT. According to our study, FVPCT accounted for approximately 68% of malignancies that were identified in FN. Because of the difference in surgical management (partial vs total thyroidectomy), efforts should be made in distinguishing between FN and FVPCT, although it has been proven that this has is a complex task.29–32 The keys to separating FN from FVPCT are the nuclear features, especially powdery chromatin and oval-shaped nuclei33 (Fig. 4). Distinguishing between FA and FC is based solely on the demonstration of histologic evidence of capsular and/or vascular invasion.25 No reliable cytologic features distinguish FA from FC.8, 34 Based on limited experience, we find that FC smears usually are more cellular with ill-defined follicles, an increased nuclear-to-cytoplasmic ratio, and hyperchromatic nuclei. The cytologic distinction between FA and FC based on cytology alone is not recommended. Differential diagnosis between Hashimoto thyroiditis and lymphoma also can be difficult, depending on the extent and grade of lymphoma. Primary lymphoma of the thyroid frequently occurs in the setting of Hashimoto thyroiditis.

Figure 1.

Cystic papillary thyroid carcinoma. A few clusters of epithelioid cells are distinct from background macrophages (Papanicolaou stain; original magnification, ×100). Inset: Epithelioid cells with dense cytoplasm and vacuolated nucleus (Papanicolaou stain; original magnification, ×400).

Figure 2.

Hyperplastic thyroid nodule. A large, single-layer sheet of follicular cells folding over itself. There are some well-formed follicles at the different levels of focus (Papanicolaou stain; original magnification, ×100). Inset: A well-formed follicle with colloid inside and matrix material outside (Papanicolaou stain; original magnification, ×400).

Figure 3.

Follicular neoplasm. Follicular cells are enlarged and present as loosely cohesive follicles and isolated, intact cells (Romanowsky stain; original magnification, ×100). Inset: Follicular cells are overlapped and form microfollicles. There is no obvious matrix material associated with them (Romanowsky stain; original magnification, ×200).

Figure 4.

Follicular variant of papillary thyroid carcinoma. Tumor cells form small, single-layer sheets and various sized follicles (Papanicolaou stain; original magnification, ×40). Inset: Tumor cells with powdery chromatin (Papanicolaou stain; original magnification, ×400).

The management of patients with thyroid nodules based on the FNA diagnostic categories is controversial, although several clinical guidelines are recommended by professional societies.8, 16, 35–38 Patients with unsatisfactory specimens should undergo a repeat FNA, preferably under sonographic guidance.38, 39 The current study demonstrated that, among patients with initial unsatisfactory specimens who underwent repeat thyroid FNA, 83% had a diagnostic repeat FNA that permitted appropriate follow-up. The malignancy rate (8.5%) in repeat FNA done because of prior unsatisfactory aspirate was comparable to the malignancy rate for patients who had a diagnostic initial FNA (7.6%). For patients with negative cytology, the reasonable management is clinical follow-up. Repeat FNA should be indicated only in patients with rapid nodular growth or other clinical symptoms that warrant further evaluation. In our study, 11.7% of patients with negative cytology had at least 1 repeat FNA, and only 5 of those patients (1.7%) had a malignancy identified, all of which were papillary carcinomas, an incidence significantly lower than that observed among patients at the first screening. A recent study demonstrated that repeat thyroid FNA in patients with initial benign cytology identified significant numbers of malignancies, which made a strong argument for repeat FNA after an initial benign FNA diagnosis.40 That study, however, did not stipulate how the patients were selected for repeat FNA. The importance of clinical information in evaluating thyroid nodules cannot be overemphasized. Considering the findings that the majority of thyroid nodules are benign, that most common thyroid malignancies have indolent biologic behavior, and that thyroid cancer can be treated effectively, even in the late stage of the disease (0.5 deaths per 100,000 population),1, 41 repeat FNA on all patients with initial benign cytology is not cost-effective and should be discouraged.42

ACL represents those specimens that suggest a benign thyroid lesion but in which FN or PCT cannot be excluded completely. In the current study, 19 of 26 patients who had repeat cytology were downgraded to a negative category, and 6 were upgraded to FN, indicating that repeat cytology was able to provide a more definitive diagnosis for management purposes.39 Of 52 patients who underwent thyroidectomy, 7 patients (13.5%) had malignant tumors identified. The malignant rate was higher than that observed in patients who had thyroid nodules in the benign category (7.3%).

The management of patients with a cytologic diagnosis of FN is a complex issue, and it is influenced by many factors.43, 44 The prevalence of malignancy in this category reportedly ranges from 20% to 60% in the literature.12, 24, 27, 45, 46 In the current study, 326 patients (63%) in this category underwent surgery. Among them, 105 patients (32%) had malignant tumors (most were FVPCT). Because of the high malignancy rate, most clinical guidelines recommend surgical removal of thyroid nodules in the FN category. Some studies have indicated that, in this category, thyroid nodules that measure >2 cm in greatest dimension carry a higher risk of malignancy,28 although other investigators did not observe a relation between nodule size and malignant potential.24, 45 Our data indicated that the average size of FC was significantly larger than that of FA (Table 7); however, the mode of FC size was smaller than that of FA, indicating that significant numbers of FC actually were smaller than FA. Therefore, a smaller nodule does not correlate with a benign lesion. In addition, FN also includes some FVPCT, and it has been demonstrated that FVPCT is unrelated to nodule size.47 Some studies have suggested that repeat cytology may help in further triaging patients in this category.39 However, this may not be practical, because, once the cytologic diagnosis of FN is made, both patients and clinicians ultimately will choose surgery. Furthermore, the second diagnosis does not necessarily negate the validity of the first.

Table 7. Size of Follicular Adenomas and Follicular Carcinomas in Surgical Specimens
Histologic diagnosisNo. of patientsMean ± SD size of nodule, cmModeP
  1. SD indicates standard deviation; FA, follicular adenoma; FC, follicular carcinoma. Mode represents the tumor size (in cm) most often seen in the surgical specimens included in this study.

FA712.30 ± 1.033 
FC294.32 ± 1.992<.0001

Patients with thyroid nodules in the suspicious for malignancy category carry a high risk for thyroid cancer. In the current study, 105 of 122 patients (86%) with nodules in this category had surgical follow-up, and malignant neoplasms were identified in 68 patients (65%). Patients who had a nodules that were suspicious for PCT a yielded much higher malignancy rate (76%) compared with patients who had nodules that were suspicious for FC (24% were proven malignant). In the early part of the study, it was believed that HN and FN with nuclear pleomorphism and atypia were more suggestive of carcinoma; however, this view no longer is held, and the data from the current study support the rationale for no longer using “suspicious for FC”. Suspicious for malignancy, therefore, should be used only when malignancies other than FC are suspected.

The study data add no new information regarding the “positive for malignancy” category. False-positive results are rare (<2% in most reports). With the exception of metastatic cancer, thyroid cancers should be managed by total thyroidectomy.48, 49

In summary, the current study demonstrated that thyroid FNA is an accurate and relatively precise tool for the diagnosis of thyroid malignancy. The 6 diagnostic categories of thyroid cytology are very useful for triaging patients with thyroid nodules for clinical management.

Ancillary