The Bethesda Thyroid Fine-Needle Aspiration Classification System: Year 1 at an Academic Institution

Abstract

Background:

Fine-needle aspiration (FNA) may be the procedure of choice in the preoperative evaluation of thyroid nodules, yet it suffers as a modality both because of its inherent limitations as well as variability in its diagnostic terminology. The National Cancer Institute recently proposed a classification system. The objective of this study was to report our experience in using this new reporting system to review the distribution of diagnosis categories and to evaluate the specificity of the system based on the cytologic–histologic correlation.

Patients and Methods:

A total of 3207 thyroid nodules underwent FNA, that is, 3207 FNAs from 2468 patients were examined at our institution between January 1, 2008 and December 31, 2008. All FNAs were classified prospectively into unsatisfactory, benign, indeterminate (cells of undetermined significance), follicular neoplasm (FN), suspicious for malignancy, and positive for malignancy.

Results:

The distribution of these categories from 3207 evaluated nodules was as follows: 11.1% unsatisfactory, 73.8% benign, 3.0% indeterminate, 5.5% FN, 1.3% suspicious, and 5.2% malignant. Of the 2468 sampled patients, 378 (15%) underwent thyroidectomy. The distribution of diagnoses of patients who underwent surgery was as follows: 10% unsatisfactory, 4.6% benign, 30.3% indeterminate, 61.4% FN, 76.9% suspicious, and 77.2% malignant. There was an excellent association between the categories and in predicting benign versus malignant thyroid nodules (p < 0.0001). However, the false-negative rate cannot be calculated because only a small number of patients with benign diagnosis underwent surgery. The false-positive rate was 2.2%; all were diagnosed as suspicious cytologically. Given that only 15% of the patients underwent surgery, at this time the sensitivity of thyroid FNA for diagnosing malignant thyroid nodules cannot be calculated, nor can the sensitivity of thyroid FNA as a screening test for all neoplasms be accurately estimated. The specificity for diagnosing malignant thyroid nodules was 93%, whereas the specificity as a screening test for all neoplasms was 68%. The positive predictive values for an FN, suspicious, and positive cytologic diagnosis were 34%, 87%, and 100%, respectively.

Conclusions:

These data demonstrate that the recently proposed classification system is excellent for reporting thyroid FNAs. Each diagnostic category conveys specific risks of malignancy, which offers guidance for patient management.

Introduction

I n the United States, about 1 out of the 14 adults will present with a thyroid nodule (1,2). However, only 5% of the thyroid nodules are found to be malignant (2,3). Therefore, it is neither necessary nor feasible to surgically remove every thyroid nodule. Fine-needle aspiration (FNA) biopsy has been widely accepted as the diagnostic procedure of choice in the evaluation of patients presenting with nontoxic thyroid nodules (4,5). The primary objective is both to triage patients, that is, to determine whether or not surgical intervention is indicated, and to assist in deciding the appropriate surgical procedure when necessary (6,7). Depending on the type of lesion, thyroid FNA can be considered as either a diagnostic test or a screening tool. As a diagnostic test, the goal is to identify papillary carcinoma and other malignancies. The exceptions are follicular and Hürthle cell carcinomas (8). Cytology is unable to differentiate follicular and Hürthle cell carcinomas from their benign counterparts because it cannot establish the presence of capsular and/or vascular invasion. Thus, thyroid FNA is treated as a screening tool with regard to the detection of follicular and Hürthle cell carcinoma, that is, FNA assesses if the nodule in question possesses the risk of being a follicular or Hürthle cell carcinoma.

About 60% of the thyroid nodules are classified cytologically as benign, whereas less than 10% of the nodules are cytologically deemed malignant (9,10). The remaining 30% present findings not diagnostic of either benignancy or malignancy. Until recently, various diagnostic terminologies, including “atypical,” “indeterminate,” and “suspicious for malignancy,” were used to describe these diagnostically challenging cases (11,12). Further, until recently there were no uniform criteria established for the various diagnostic categories and specimen adequacy (13,14). As a result, diagnostic inconsistencies exist among different laboratories as well as pathologists within the same laboratories. This results in difficulty in communicating the clinical implications of thyroid FNA results both to direct caregivers (endocrinologists and surgeons) and indirect caregivers (pathologists and radiologists) (15).

It is apparent that there is a need for a consensus on the diagnostic terminologies and the use of consistent criteria for determination of specimen adequacy and diagnostic categories such that the direct caregivers can readily understand the prognostic meaning of the results of thyroid FNA. In an attempt to establish a standardized diagnostic terminology/classification system and morphologic criteria for reporting thyroid FNA, the National Cancer Institute (NCI) sponsored the NCI Thyroid FNA State of the Science Conference, an interdisciplinary program that included a permanent website and 2-day meeting, which convened a group of experts in thyroid FNA at Bethesda, MD, in October 2007 (16). One of the recommendations endorsed by this program was the establishment of a six-tiered diagnostic classification system based on a probabilistic approach (17).

On January 1, 2008, our institution adopted this six-category diagnostic approach in reporting thyroid FNA with minor modifications. This prospective study reports our experience employing this system for thyroid FNA in a 12-month period, including the incidence and histologic outcomes of each diagnostic category. We also calculated the operating characteristic of this new diagnostic approach in identifying neoplasms and malignancies.

Patients and Methods

Study population

The Institutional Review Board-approved (#8219) study population included all 3207 consecutive thyroid FNAs that were evaluated at the Yale-New Haven Hospital from January 1, 2008 to December 31, 2008. Either endocrinologists or radiologists performed all of the biopsies. The majority of the FNAs were performed under ultrasound guidance. Direct smears were prepared from each pass and were either air dried and stained with the Diff-Quik stain (Dade Behring, Deerfield, IL) or fixed in alcohol and stained by the Papanicolaou method. The needle was then rinsed in Cytolyt solution (Hologic, Marlborough, MA) for ThinPrep preparation. On average, two to three passes were performed for each nodule. On-site assessment of adequacy was performed in 2.5% of the patients. The study cases also included specimens submitted by outside laboratories for second opinion. These outside cases consisted of direct smears and/or ThinPrep preparation; the former could be using Diff-Quik stain, Papanicolaou method, or hematoxylin–eosin.

Cytologic diagnostic terminology and interpretation

At our institution, a six-tier diagnostic approach was developed according to the recommendations of 2007 NCI Thyroid FNA State of the Science Conference with minor modifications (17). In-service training was given to all cytotechnologists and cytopathologists to explain the terms and implications of each diagnostic category before implementing the new diagnostic approach on the first day of 2008. An educational session explaining the new diagnostic approach was also offered to the endocrinologists, surgeons, and radiologists.

The six-tier diagnostic approach included the following categories: unsatisfactory, benign, indeterminate (cells of undetermined significance), follicular neoplasm (FN), suspicious for malignancy, and positive for malignancy.

Unsatisfactory

These specimens demonstrated inadequate cellularity, poor fixation and preservation, obscuring blood or ultrasound gel, or a combination of these factors. Inadequate cellularity was defined as the presence of less than six groups of well-preserved follicular cells on each of at least two slides. Aspirates from nodules that demonstrated abundant colloid and/or macrophages with little or no follicular cells were not considered unsatisfactory if these findings correlated with the clinical and radiologic impression of a colloid nodule or cystic lesion.

Benign/negative for malignancy

This category included the diagnoses of nodular goiter, nodular goiter with hyperplastic nodules, colloid nodules, cyst contents with or without benign follicular cells, and lymphocytic thyroiditis.

Indeterminate

This category was equivalent to the category of follicular cells of undetermined significance in the classification system proposed by the NCI Thyroid FNA State of the Science Conference. This category was reserved for aspirates with findings that were not convincingly benign but were not diagnostic of a neoplastic or malignant process. These cases were often characterized by borderline cellularity. We further subdivided this category into two microscopic descriptive subcategories. One subcategory was for cases that demonstrated a predominant microfollicular pattern and no or minimal colloid, that is, an architectural pattern eliciting concern for a FN. The second subcategory was for cases that demonstrated nuclear atypia such as the presence of occasional nuclear grooves, abnormal chromatin pattern, or nuclear overlapping and crowding, that is, anomalous findings within individual cells eliciting concern for a papillary carcinoma. A comment was attached to these diagnoses alerting the direct caregiver that the term corresponds to the NCI-proposed category cells of undetermined significance and that this subset of patients may benefit from repeat FNA in the appropriate clinical context.

Follicular neoplasm

This category applied to cellular specimens with abundant follicular cells arranged in a microfollicular pattern with little or no colloid. The differential diagnosis included hyperplastic adenomatoid nodules, follicular adenoma, follicular carcinoma, and follicular variant of papillary carcinoma where the nuclear features remain ill defined. For cases that demonstrated, in addition to microfollicles, some but not all of the nuclear features of papillary carcinoma, the comment, “suggestive of Follicular Variant of Papillary Carcinoma” was included in the report. The FN category also included cases that demonstrated a predominant population of Hürthle cells; these cases would be labeled Hürthle cell neoplasm. The differential diagnosis would include hyperplastic adenomatoid nodule with Hürthle cell change, Hürthle cell adenoma, and Hürthle cell carcinoma.

Suspicious for malignancy

This category applied to the specimens that demonstrated features of a malignant neoplasm but were quantitatively or qualitatively insufficient to make a definitive diagnosis of malignancy. These features included, but were not limited to, an occasional intranuclear inclusion, nuclear grooves, or psammoma calcifications (calcospherites).

Positive for malignancy

This category applied to cellular specimens with unequivocal cytologic evidence of a malignant neoplasm. The category included papillary carcinoma and its variants, medullary carcinoma, anaplastic carcinoma, lymphoma, and metastatic lesions.

The cytologic diagnoses were rendered by five board-certified cytopathologists; their experience ranged from 3 to 10 years. All cases that were classified as FN, suspicious of malignancy, or positive for malignancy as well as a substantial number of indeterminate and negative cases were reviewed in the daily consensus conference among the cytopathologists.

Follow-up

Histologic diagnoses of patients who had undergone surgery were used as the gold standard for correlation with the cytologic interpretations. In the event where more than one nodule underwent biopsy on the same patient, the most abnormal FNA result was used for analysis. The specimens with discrepant cytologic and histologic diagnoses were reviewed to determine the plausible explanations of these discrepancies.

Statistical analysis

Categorical analysis was performed using both a log-linear model (likelihood ratio) and χ ² model with symmetric measures of association, using a two-sided p-value of <0.05 as significant (SPSS 16.0; SPSS, Chicago, IL). The specificity was estimated using two approaches: (i) as a diagnostic test—the performance of thyroid FNA to identify malignancy in which FNA specimens that were interpreted as suspicious of malignancy or positive for malignancy were classified as positive and the remaining categories classified as negative and (ii) as a screening test—the performance of thyroid FNA to identify neoplasms in which specimens in the benign/negative for malignancy category were considered negative and the remaining categories considered positive. Cases that were classified as unsatisfactory and indeterminate were excluded from calculations for both approaches inasmuch these categories communicate the absence of diagnostic material and the need for additional sampling rather than the presence of neoplasm/malignancy. Because it was not possible to distinguish follicular or Hürthle cell adenomas from their malignant counterparts cytologically, cases that were classified as FN were excluded from calculation for the first approach, that is, excluded from assessment of FNA as a diagnostic test.

Results

A total of 3207 thyroid FNAs obtained from 2468 patients were evaluated at our institution during the 12-month study period. Two hundred and seventy-one (11%) patients were aspirates submitted by outside laboratories for a second opinion. Table 1 summarizes the distributions of cytologic diagnoses both according to the nodules and patients. Overall, 378 (15%) patients underwent surgical resection that included either partial or total thyroidectomy. Table 2 summarizes the cytologic–histologic correlation.

Table 1.

Cytologic Diagnosis by Nodules and Patients

Cytologic category	By nodules	By patients
Unsatisfactory	357 (11.1%)	230 (9.3%)
Benign/negative for malignancy	2368 (73.8%)	1799 (72.9%)
Indeterminate/cells of undetermined significance	95 (3.0%)	89 (3.6%)
FN/Hürthle cell neoplasm	176 (5.5%)	166 (6.7%)
Suspicious for malignancy	43 (1.3%)	39 (1.6%)
Malignancy	168 (5.2%)	145 (5.9%)
Total	3207	2468

FN, follicular neoplasm.

Table 2.

Patients with Cytologic and Histologic Correlation

	Histologic follow-up
			Malignant neoplasm
Cytologic category	MNG/HT	FA	PTC	FC	Others	Total	Total
Unsatisfactory	9	8	7	1	0	8	25
Benign/negative for malignancy	61	13	8	0	0	8	82
Indeterminate/cells of undetermined significance	7	7	11	2	0	13	27
FN/Hürthle cell neoplasm	33	34	23	12	0	35	102
Suspicious for malignancy	2	2	26	0	0	26	30
Malignancy	0	0	100	1	11	112	112
Total	112	64	175	16	11	202	378

MNG, multinodular goiter; HT, Hashimoto thyroiditis; FA, follicular and Hürthle cell adenoma; PTC, papillary thyroid carcinoma; FC, follicular or Hürthle cell carcinoma. “Others” indicate medullary carcinoma, anaplastic carcinoma, and metastatic lesions (specifically squamous cell carcinoma of the head and neck).

Of the 230 patients with an unsatisfactory diagnosis, 34 patients had repeat FNA. The diagnoses of repeat FNA included 14 unsatisfactory, 19 negative for malignancy, and 1 positive for papillary carcinoma. Twenty-five of the 230 (10.9%) patients with an unsatisfactory FNA underwent surgical resection; 9 patients were found to have benign goiters, 8 with follicular adenomas, and 7 with papillary carcinomas.

In all, 2386 of the 3207 (73.8%) thyroid nodules were interpreted cytologically as negative for malignancy; the majority of them were nodular goiters, followed by lymphocytic thyroiditis and colloid nodules. Eighty-two of the 2468 (3.3%) patients underwent thyroidectomy because of other clinical considerations such as the size of the nodules, presence of family history of thyroid malignancy, and/or prior history of irradiation to the neck. The 82 patients exhibited histologic diagnoses that included 61 (74.3%) nodular goiters, 13 (15.9%) follicular adenomas, and 8 (9.7%) papillary thyroid carcinomas (PTC). All of the PTCs that were discovered histologically in patients with benign cytologic diagnoses were 1.0 cm or less; further, six of them were 0.5 cm or less. The targeted nodules included a 3.6 cm hemorrhagic cyst with an adjacent 0.4 cm nontargeted PTC, 1.3, 1.5, 2.0, and 4.1 cm dominant nodules within four patients' goiters containing 0.2, 1.0, 0.1, and 1.0 cm PTCs, respectively. In addition, a 3.4 cm dominant nodule within a Hashimoto thyroiditis patient contained a nontargeted 0.5 cm PTC. Two patients were consults that did not include the sizes of the targeted nodules yet contained multifocal papillary microcarcinomas (dimensions ≤0.5 cm). None of these cases demonstrated any cytologic evidence of papillary carcinoma on retrospective review of cytologic material.

Eighty-nine (3.6%) of the 2468 patients who underwent FNA were interpreted as indeterminate (cells of undetermined significance); of these 89 patients, 58 (65.2%) were further subclassified as low cellularity with microfollicular architecture and absence of colloid and 31 (34.8%) demonstrated nuclear features not characteristic of benign lesions (i.e., nuclear atypia). Seventeen of the 89 (19%) patients had a repeat thyroid FNA. The repeat FNA diagnoses included 3 unsatisfactory, 11 negative for malignancy, 1 indeterminate, 1 FN, and 1 positive for papillary carcinoma. Twenty-seven of the 89 (30.3%) patients underwent thyroidectomy, 7 were found to have nodular goiters, 7 follicular adenomas, and 13 carcinomas, including 2 follicular carcinomas and 11 papillary carcinomas. Interestingly, all 11 cases that were found to be papillary carcinoma on subsequent surgical resection had preoperatively demonstrated cytologic nuclear atypia.

One hundred and seventy-six (5.5%) of the 3207 thyroid nodules evaluated and corresponding to 166 (6.7%) of the 2468 patients were diagnosed as FN. These included 67 of the 176 (38.0%) nodules from 63 of the 166 (37.9%) patients that were diagnosed as Hürthle cell neoplasm. One hundred and two (61.4%) of the 166 patients underwent surgery. The 39.6% who were not operated upon at the time of this study were either followed outside our institution or are currently being followed and are likely to undergo surgery. The histologic findings included 33 (32.4%) of the 102 patients with nodular goiters, 34 (33.3%) with follicular adenomas, and 35 (34.3%) with carcinomas. The latter included 12 follicular carcinomas and 23 papillary carcinomas. Among all the papillary carcinomas diagnosed on histology, 23 of the 102 (22.5%) were found to be the follicular variant; in addition, 3 of which were interpreted as FN with features suggestive of follicular variant of papillary carcinoma cytologically. Therefore, FN at our institution has a one-third risk of being malignant and an increased likelihood of being a follicular carcinoma or a follicular variant papillary carcinoma.

A total of 43 (1.3%) FNAs of the 3207 sampled nodules from 39 (1.6%) of the 2468 patients were diagnosed as suspicious for malignancy; all were suspicious for papillary carcinoma. Thirty (77%) of the 2468 patients underwent surgical resection, which showed 4 false positives: 2 (6.7%) of the 30 patients exhibiting nodular goiters with papillary hyperplasia and 2 (6.7%) follicular adenomas. The remaining 26 (86.7%) of the 30 patients showed PTCs.

One hundred and sixty-eight (5.2%) of the 3207 sampled thyroid nodules from 145 (5.9%) of the 2468 patients were classified as positive for malignancy cytologically: 128 patients had a cytologic diagnosis of papillary carcinoma, 4 medullary carcinomas, 3 anaplastic carcinomas, 1 diffuse large B cell non-Hodgkin lymphoma, 1 Hürthle cell carcinoma, and 3 metastatic lesions, all of which were squamous cell carcinomas of the head and neck. One hundred and twelve (77.2%) of the 128 patients with positive cytology underwent thyroidectomy at the time of this study; the histologic diagnoses included 100 papillary carcinomas, 1 follicular carcinoma, 4 medullary carcinomas, 3 anaplastic carcinomas, 1 Hürthle cell carcinoma, and 3 metastatic squamous cell carcinomas. No false positives were identified in the positive for malignancy diagnostic category.

Overall, there were excellent correlations between the six FNA diagnostic categories and the histologic outcomes in predicting nonneoplastic versus neoplastic thyroid nodules and benign versus malignant thyroid nodules (Table 3). Comparing each individual cytologic diagnostic category against the other four categories revealed statistically significant differences between benign and indeterminate, benign and FN, benign and suspicious, benign and malignant, indeterminate and suspicious, indeterminate and malignant, FN and suspicious, FN and malignant, as well as suspicious and malignant. However, there was no statistically significant difference between indeterminate and FN.

Table 3.

Statistical Analysis of the Cytologic Classification System

Comparisons between different cytology categories	χ² ^a	Phi ^b	LR ^c	p-Value (two-sided)
All categories	224	0.891	244:1	<0.001
All categories excluding the unsatisfactory category	224	0.908	243:1	<0.001
B vs. I	25	0.574	23:1	<0.001
B vs. FN	43	0.571	47:1	<0.001
B vs. S	47	0.749	48:1	<0.001
B vs. M	140	0.934	171:1	<0.001
I vs. FN	1	0.103	1:1	NS
I vs. S	7	0.414	7:1	0.035
I vs. M	61	0.725	41:1	<0.001
FN vs. S	96	1.000	106:1	<0.001
FN vs. M	19	0.441	20:1	<0.001
S vs. M	96	0.746	117:1	<0.001

χ ² model for association between categorical data.

Phi correlation coefficient for measuring the strength of association between categorical data; values range from +1 to −1; +1 denotes strong positive association, −1 strong negative association, and 0 no association.

Likelihood ratio (log-linear model) for association between categorical data.

B, benign/negative for malignancy; I, indeterminate; S, suspicious for malignancy; M, positive for malignancy; NS, not statistically significant.

Based on the cytologic–histologic correlation of the 378 patients, there were 4 false-positive diagnoses. We calculate the specificity of thyroid FNA as a diagnostic test, that is, in predicting a malignant process, to be 93% (Table 4). Only 15% of the 2468 patients underwent surgery, and therefore, a true sensitivity cannot be assessed. In addition, the negative predictive value of a benign/negative for malignancy cytologic interpretation cannot be calculated. We cannot justifiably consider the small number of micropapillary carcinomas as false negatives, nor can we calculate the false-negative diagnostic rate because of the low number of benign patients who underwent surgery. Given the above, we calculate that the specificity of predicting a malignancy is 93%. The positive predictive value of a malignant and suspicious cytologic diagnosis was 100% and 83%, respectively, and a positive predictive value of 97% if we combined both categories into one.

Table 4.

Operating Characteristic of Thyroid Fine-Needle Aspiration

	As a screening test for neoplasm	As a diagnostic test for malignancy
Sensitivity	Cannot be calculated^a	Cannot be calculated^a
Specificity	68%	93%
Positive predictive value	88%	97%
Negative predictive value	Cannot be calculated^a	Cannot be calculated^a

Given the selection bias of only 82 patients diagnosed as benign undergoing surgery, sensitivity, false-negative rate, and negative predictive values cannot be accurately calculated.

We calculate the specificity of thyroid FNA as a screening test, that is, in predicting a neoplastic process, to be 68%. Given that only 15% of the 2468 patients underwent surgery, a true sensitivity therefore cannot be assessed. Nor can the negative predictive value of a negative cytologic diagnosis for a neoplastic process be calculated. The positive predictive value of a cytologic interpretation as a screening test for a neoplastic process, that is, FN, suspicious, and malignant, is 88%.

Discussion

The current study attempted to evaluate the efficacy of the newly proposed six-tier diagnostic classification system in reporting thyroid FNA results. Based on a cohort of 3207 thyroid FNAs, 11.1% of the thyroid nodules were classified as unsatisfactory, 73.8% benign, 3.0% indeterminate/cells of undetermined significance, 5.5% FNs, 1.3% suspicious for malignancy, and 5.2% positive for malignancy. In a recent study that also utilized a similar six-tier cytologic diagnostic approach, the authors reported that 10.4% were classified as unsatisfactory, 64.6% were classified as benign, 3.2% were classified as atypical cells/lesion, 11.6% were classified as FN, 2.6% were classified as suspicious, and 7.6% were classified as malignant based on a cohort of 4073 thyroid FNA samples from two academic institutions (10). Although we observed a lower proportion of cases diagnosed as FN, suspicious, and positive for malignancy, there was no statistically significant difference among the distribution of the diagnostic categories between the two studies (paired t-test, p > 005). If we combined both indeterminate and the FN categories into a single diagnostic category, the frequency of distribution of the diagnostic categories was similar to those published in a recent report that utilized a five-tier diagnostic approach (unsatisfactory 5% vs. 11.7%, benign 75% vs. 78%, indeterminate 13% vs. 8.8%, suspicious 3% vs. 1.4%, and malignant 4% vs. 5.5%) (18).

One of the major limitations of thyroid FNA is the relatively high rate of unsatisfactory/nondiagnostic specimens. The majority of the reported unsatisfactory rates ranged between 10% and 20% (10,18,19). However, unsatisfactory rates as high as 30% have been reported (9,20). The rate of nondiagnostic specimens may be influenced by the nature of the thyroid nodules, the experience of the aspirators, whether on site adequacy assessment is performed, and the criteria used to define specimen adequacy. Although a nondiagnostic category was part of the diagnostic scheme recommended by the NCI Thyroid FNA State of the Science Conference, the criteria for defining specimen adequacy were not explicitly spelled out. In our institution, we used the number of follicular cells as a general rule but not the absolute criterion. In many instances, we took into consideration the sonographic findings. For nodules that appeared entirely cystic or colloid rich on ultrasound with the aspirate demonstrating predominantly histiocytes and/or colloid, the specimen would be classified as adequate even if there were an insufficient number of follicular cells; a qualifier indicating paucity or lack of follicular cells was often added. This approach proved useful because our unsatisfactory rate was relatively low, less than 12%. Another plausible explanation was that most of our FNA were performed under ultrasound guidance. In lesions greater than 1.5 cm, the diagnostic yield of ultrasound-guided FNA is modestly superior to conventional FNA, whereas in lesions less than 1.5 cm, ultrasound guidance may offer its greatest utility (21 –23). In this study, approximately 11% of the 3207 nodules that underwent FNA with initial nondiagnostic interpretations were followed by repeat FNA; the majority of them had a diagnostic aspirate after repeat FNA. Last but not least, on-site cytologic specimen adequacy assessments have been shown to significantly decrease unsatisfactory rates (24,25). Accordingly, we plan to invigorate our on-site specimen adequacy assessment capacity.

It is not surprising that aspirates classified as benign/negative for malignancy accounted for the majority of the thyroid FNAs. This category consisted of a heterogeneous group of lesions such as nodular goiter, colloid nodule, and lymphocytic thyroiditis. Most of these patients were followed up clinically. However, a small number, 82 (4.6%) of the 1799 patients with a benign cytologic interpretation, underwent surgery because of the size of the nodules, presence of a family history of thyroid cancer, or history of irradiation to the head and neck area, that is, these patients had higher risk of developing or harboring a thyroid malignancy than the general population. Eight (9.8%) of the 82 patients who had a histologic follow-up were to found have micro PTCs. Given the small sample of patients who underwent surgery and the selection bias inherent in them being operated upon, we cannot justifiably calculate a false-negative rate for our institution. Typically, the reported rates of false-negative cytology diagnoses reported in the literature are rather low (9,10,19,26,27). Based on the same rationale, an assessment of the risk of malignancy of a benign diagnosis cannot be made (Table 5). On secondary review, none of the cytology aspirates that contained micro PTCs on final histology demonstrated any evidence of papillary carcinoma. Only four of the missed cancers represented errors of sampling within the nodule targeted by FNA. The remaining missed cancers represented nontargeted micropapillary carcinomas. In addition, micropapillary carcinoma is often considered as an incidental finding (28,29). If these micropapillary carcinomas were not counted as misses, there are no false-negative cases in our current series. In addition, 13 patients were found to have a follicular adenoma on histologic follow-up. Cytologically, it is impossible to distinguish between follicular adenoma and hyperplastic adenomatous nodule. We do not consider these cases to be discrepancies because there is only minimal risk to the patients.

Table 5.

Risk of Malignancy per Diagnosis

Thyroid FNA cytologic diagnosis (% of patients that went to surgery)	Percent malignant on histology [NCI recommended rate ^b ]
Benign/negative for malignancy (3% or 82/2468 patients)	9.8% (8 of the 82)^a [No NCI rate given]
Indeterminate/cells of undetermined significance (30% or 27/89 patients)	48% (13/27)^a [10–15%]
FN (61% or 102/166 patients)	34% (35/102) [20–30%]
Suspicious for malignancy (77% or 30/39 patients)	87% (26/30) [75%]
Positive for malignancy (77% or 112/145 patients)	100% (112/112) [100%]

These values may in fact be lower given that the subset of patients with these diagnoses that underwent surgery demonstrated increased clinical risk of malignancy.

The NCI Guidelines on Thyroid FNA include recommendations as to an expected/acceptable range of malignancy risk per diagnostic category.

FNA, fine-needle aspiration; NCI, National Cancer Institute.

The diagnostic category indeterminate included cases that could not be classified as benign or FN and was equivalent to the diagnostic category follicular lesion of undetermined significance/atypia of undetermined significance in the NCI classification scheme. The guidelines provided by the NCI encouraged minimizing the use of this category. In our study, the indeterminate category accounted for only 3% of all thyroid FNAs, the second lowest in frequency among all diagnostic categories, only preceded by the suspicious for malignancy category. According to the NCI recommendation, patients with an indeterminate thyroid FNA could benefit from repeat FNA because the risk of malignancy was 5–10% (11,17). Surprisingly, only 19% of our patients underwent repeat FNA. One possible explanation is unfamiliarity with the NCI recommendations. In view of this possibility, the following comment “At our institution, the term ‘Indeterminate’ corresponds to the NCI recommended diagnostic classification ‘Follicular Cells of Undetermined Significance’. This subset of patients may benefit from repeat FNA in the appropriate clinical context” has been routinely included in the report of indeterminate thyroid FNAs.

Only 27 (30%) of our patients with an indeterminate cytologic interpretation underwent thyroidectomy. Thirteen malignancies, including 2 follicular carcinomas and 11 papillary carcinomas, were identified on subsequent histology. Based only on cases with histologic follow-up, almost 50% of the patients turned out to be malignant. However, this did not represent the actual risk of malignancy for patients with an indeterminate FNA diagnosis. It would be impossible to accurately estimate the true risk of malignancy after an indeterminate diagnosis because not all indeterminate thyroid nodules were resected.

We routinely classified indeterminate FNAs into two subcategories, cases with borderline/low cellularity with microfollicular pattern and those with nuclear atypia. The latter referred to the presence of nuclear enlargement, mild nuclear overlapping and crowding, abnormal chromatin pattern, and/or nuclear grooves; intranuclear inclusions were usually not identified. All 11 thyroid FNAs that were found to have papillary carcinomas on subsequent follow-up demonstrated nuclear atypia. Although the number of cases was small, our observation suggests that the presence of nuclear atypia might be an independent risk factor for patients with indeterminate FNAs. Kelman et al. also noticed that 60% of the indeterminate thyroid nodules demonstrating nuclear atypia cytologically were found to be malignant, whereas only 7% of the indeterminate thyroid nodules without nuclear atypia were proven to be malignant (30). Our observations and others support the notion that further subclassification of the indeterminate category according to the presence or absence of nuclear atypia may provide additional predictive value. Further, despite the relative immaturity of the data regarding the indeterminate category, a presumptive institutional philosophy has arisen with regard to this diagnosis. Currently, we believe that nodules with an indeterminate/cells of undetermined significance FNA diagnosis should be rebiopsied and not be surgically resected. Given the often-borderline specimen adequacy of indeterminate samples, we consider a targeted, well-sampled re-FNA with a diagnosis of benignity to be a reassuring finding indicative of conservative management. We believe that an interpretation of indeterminate on a re-FNA may justify surgical intervention in the appropriate clinical circumstances based on the rationale that repeat finding of indeterminate suggests that rather than being a sampling issue, such diagnosis may reflect a relatively underdeveloped atypia within the lesion itself, especially in cases exhibiting nuclear atypia. Re-FNAs with FN, suspicious, and positive diagnoses should be managed according to the standard protocols.

The diagnostic category FN was applied to cellular aspirates that demonstrated a predominant microfollicular pattern with little or no colloid. Because of the inability of FNA to distinguish follicular and Hürthle cell adenomas from their malignant counterparts, the role of thyroid FNA shifts from being a diagnostic test to a screening test, implying that specificity will take a back seat with regard to sensitivity in detecting follicular and Hürthle cell carcinoma. Because of the immediate risk of malignancy, it is not surprising that a majority of the patients with this diagnosis underwent thyroid lobectomy. About one-third of our patients were found to have nonneoplastic goiters, one-third had adenomas, and another one-third had a malignancy. These observations were consistent with the expected prevalence of nonneoplastic (∼25%) and malignant (20–30%) nodules after diagnosis of FN (11,17,31,32). All 23 histologically proven papillary carcinomas demonstrated features characteristic of follicular variant; most of the aspirates demonstrated variable degree of nuclear atypia in addition to a microfollicular pattern. Thus, our institutional philosophy for patients with either follicular or Hürthle cell neoplasms is that surgical intervention is appropriate. If a repeat biopsy was to demonstrate less concerning findings such as a typical hyperplastic nodule, that result, in and of itself, would not obviate the previous finding, given that a good quality specimen is the basis of patient management. Accordingly, surgical intervention is always based on the most, rather than the least, concerning cytologic findings.

In our institution, we further separated the category of FN into those demonstrating predominantly follicular cells and those demonstrating predominantly Hürthle cells because some authors have indicated that the risk of malignancy was higher in Hürthle cell lesions, especially when the nodule was 3.5 cm or larger (17). In the current study, 9 of the 37 (24%) Hürthle cell neoplasm and 26 of the 64 (41%) follicular cell neoplasm with histologic follow-up turned out to be malignant; the difference was statistically significant. A recent study has also observed that there was no statistically significant difference in the frequency of cancer between nodules with a cytologic diagnosis of FN and those with a diagnosis of Hürthle cell neoplasm (31% vs. 28%) (33). Our observation and others did not support the notion that the risk of malignancy was higher in patients with a cytologic diagnosis of Hürthle cell neoplasm than those with a cytologic diagnosis of FN. The sizes of the nine Hürthle cell neoplasms ranged from subcentimeter to over 3.5 cm without any size association to malignant potential.

It is generally agreed that an equivocal diagnostic category is appropriate and helpful in a situation in which there is diagnostic uncertainty. Several studies using a five-tier diagnostic approach have one diagnostic category to represent diagnostic uncertainty (18,26,34). In the current study, we followed the NCI's recommendation and used two diagnostic categories indeterminate and FN to denote diagnostic uncertainty. The former category consisted of cases that were of low/borderline cellularity, had a relatively low risk of malignancy (5–10%), and would benefit from repeat FNA. The latter category was designated for cases that demonstrated an overlap in cytologic features of adenomatoid nodule, FN, and follicular carcinoma. Because of the substantial risk of malignancy (25–33%), patients with a cytologic diagnosis of FN would benefit from immediate surgical intervention. We did not observe any significant difference in the histologic outcomes between aspirates with a cytologic diagnosis of indeterminate and those with a cytologic diagnosis of FN. The most likely explanation is that the indeterminate nodules that underwent surgery were a selected group with increased clinical risk of harboring a thyroid malignancy; we did not include those nodules that demonstrated low clinical risk, and were therefore managed conservatively, into the calculation. In our opinion, the use of two equivocal diagnostic categories allows the reporting cytopathologists to express their level of concern of the possibility of an underlying malignancy to guide subsequent patient management.

Forty-three (1.3%) and 168 (5.2%) of the 3207 thyroid nodules that underwent FNA had a cytologic diagnosis of suspicious for malignancy and positive for malignancy, respectively. The majority of these cases were suspicious or positive for papillary carcinoma specifically. Three-quarters of patients with these diagnoses underwent thyroidectomy. Histologic follow-up did not reveal any false positives for cases that were interpreted as definitively malignant cytologically, that is, there were no false positives in the positive for malignancy category. However, there were four false positives for cases that were interpreted as suspicious for malignancy cytologically; they included two follicular adenomas and two papillary hyperplastic nodules. The latter entity has been recently recognized as a potential pitfall in the cytologic diagnosis of papillary carcinoma (35). Nonbranching papillary groups can be seen in both conditions; the presence of focal oncocytic changes and the absence of convincing nuclear atypia of PTC favor papillary hyperplastic nodule. The two follicular adenomas were associated with sclerosis, extensive degeneration, and occasional nuclear features that mimicked those of papillary carcinoma. Further, with regard to the suspicious category, 3 of the 26 patients with suspicious cytology and eventual malignant histology underwent lobectomy with subsequent completion thyroidectomy. Two of the patients who had been diagnosed as suspicious were themselves status postlobectomy for PTC in the past and were currently receiving completion thyroidectomy. The remaining 21 (80.7%) of the 26 patients with suspicious cytology and PTC on subsequent histology had undergone total thyroidectomy initially, often with intraoperative frozen section consultation. Thus, one can argue that there was no major adverse impact on patient care because all of the patients with eventual malignant histology underwent surgery with the majority (80.7%) of these patients receiving total thyroidectomy in the first place. The risk of malignancy for a diagnosis of suspicious for malignancy and positive for malignancy was 87% and 100%, respectively, consistent with the data, 75% and 100%, respectively, published in the literature (Table 5) (17). The relative high incidence of malignant lesions after a suspicious diagnosis suggests that the criteria of malignant category may be too restrictive and that some of these patients, in retrospect, could have been categorized as outright positive for papillary carcinoma.

In summary, our study shows that the six-tier diagnostic approach of reporting thyroid FNA proposed by the NCI Thyroid FNA Scientific Conference is an excellent screening test in identifying patients who may harbor a follicular or Hürthle cell neoplasm with an acceptable specificity (68%). Using the same diagnostic approach, we have also shown that thyroid FNA is a superb diagnostic test in identifying papillary carcinomas and other malignancies with a specificity of 93%. By adhering to strict cytologic criteria, we encountered few false positives. Further, each diagnostic category conveys a different level of risk of malignancy to the direct caregivers. Given that the vast majority of patients with a cytologic interpretation of negative for malignancy did not undergo surgical follow-up, we cannot speculate as to its exact risk of malignancy but must rely on standard conservative clinical monitoring unless surgical intervention is indicated because of other clinical/radiological considerations. On the other hand, thyroid surgery is indicated for patients with a cytologic interpretation of FN, suspicious for malignancy, and positive for malignancy because of the substantial risk of malignancy. For patients with an indeterminate thyroid FNA, benefit may be obtained from repeat thyroid FNA in the absence of clinical and/or radiologic concerns of an underlying malignancy.

Footnotes

Disclosure Statement

Constantine G.A. Theoharis: no competing financial interests exist.

Kevin M. Schofield: no competing financial interests exist.

Lynwood Hammers: Dr. Hammers is Medical Director, Ultrasound Specialties, Hammers Healthcare Imaging of New Haven, CT.

Robert Udelsman: no competing financial interests exist.

David C. Chhieng: no competing financial interests exist.

References

Vander

, Gaston

, Dawber

. 1968. The significance of nontoxic thyroid nodules. Final report of a 15-year study of the incidence of thyroid malignancy. Ann Intern Med, 69:537–540.

Rojeski

, Gharib

. 1985. Nodular thyroid disease. Evaluation and management. N Engl J Med, 313:428–436.

Mazzaferri

. 1992. Thyroid cancer in thyroid nodules: finding a needle in the haystack. Am J Med, 93:359–362.

Mazzaferri

. 1993. Management of a solitary thyroid nodule. N Engl J Med, 328:553–559.

Cooper

, Doherty

, Haugen

, Kloos

, Lee

, Mandel

, Mazzaferri

, McIver

, Sherman

, Tuttle

. The American Thyroid Association Guidelines Taskforce. 2006. Management guidelines for patients with thyroid nodules and differentiated thyroid cancer. Thyroid, 16:109–142.

Gharib

. 1994. Fine-needle aspiration biopsy of thyroid nodules: advantages, limitations, and effect. Mayo Clin Proc, 69:44–49.

Zagorianakou

, Malamou-Mitsi

, Zagorianakou

, Stefanou

, Tsatsoulis

, Agnantis

. 2005. The role of fine-needle aspiration biopsy in the management of patients with thyroid nodules. In Vivo, 19:605–609.

Werga

, Wallin

, Skoog

, Hamberger

. 2000. Expanding role of fine-needle aspiration cytology in thyroid diagnosis and management. World J Surg, 24:907–912.

Amrikachi

, Ramzy

, Rubenfeld

, Wheeler

. 2001. Accuracy of fine-needle aspiration of thyroid. Arch Pathol Lab Med, 125:484–488.

10.

Yang

, Schnadig

, Logrono

, Wasserman

. 2007. Fine-needle aspiration of thyroid nodules: a study of 4703 patients with histologic and clinical correlations. Cancer, 111:306–315.

11.

The Papanicolaou Society of Cytopathology Task Force on Standards of Practice. 1997. Guidelines of the Papanicolaou Society of Cytopathology for fine-needle aspiration procedure and reporting. Diagn Cytopathol, 17:239–247.

12.

Baloch

, LiVolsi

. 2004. Fine-needle aspiration of thyroid nodules: past, present, and future. Endocr Pract, 10:234–241.

13.

Eedes

, Wang

. 2004. Cost-effectiveness of immediate specimen adequacy assessment of thyroid fine-needle aspirations. Am J Clin Pathol, 121:64–69.

14.

Kelly

, Lim

, DeJong

, Harmath

, Dudiak

, Wojcik

. 2006. Specimen adequacy and diagnostic specificity of ultrasound-guided fine needle aspirations of nonpalpable thyroid nodules. Diagn Cytopathol, 34:188–190.

15.

Redman

, Yoder

, Massoll

. 2006. Perceptions of diagnostic terminology and cytopathologic reporting of fine-needle aspiration biopsies of thyroid nodules: a survey of clinicians and pathologists. Thyroid, 16:1003–1008.

16.

Abati

. 2008. The National Cancer Institute Thyroid FNA state of the science conference: “wrapped up” Diagn Cytopathol, 36:388–389.

17.

Baloch

, Cibas

, Clark

, Layfield

, Ljung

, Pitman

, Abati

. 2008. The National Cancer Institute Thyroid fine needle aspiration state of the science conference: a summation. Cytojournal, 5:1–17.

18.

Yoder

, Redman

, Massoll

. 2006. Validation of a five-tier cytodiagnostic system for thyroid fine needle aspiration biopsies using cytohistologic correlation. Thyroid, 16:781–786.

19.

Gharib

, Goellner

, Johnson

. 1993. Fine-needle aspiration cytology of the thyroid. A 12-year experience with 11,000 biopsies. Clin Lab Med, 13:699–709.

20.

Burch

. 1995. Evaluation and management of the solid thyroid nodule. Endocrinol Metab Clin North Am, 24:663–710.

21.

Redman

, Zalaznick

, Mazzaferri

, Massoll

. 2006. The impact of assessing specimen adequacy and number of needle passes for fine-needle aspiration biopsy of thyroid nodules. Thyroid, 16:55–60.

22.

Danese

, Sciacchitano

, Farsetti

, Andreoli

, Pontecorvi

. 1998. Diagnostic accuracy of conventional versus sonography-guided fine-needle aspiration biopsy of thyroid nodules. Thyroid, 8:15–21.

23.

Cesur

, Corapcioglu

, Bulut

, Gursoy

, Yilmaz

, Erdogan

, Kamel

. 2006. Comparison of palpation-guided fine-needle aspiration biopsy to ultrasound-guided fine-needle aspiration in the evaluation of thyroid nodules. Thyroid, 16:555–561.

24.

Nasuti

, Gupta

, Baloch

. 2002. Diagnostic value and cost-effectiveness of on-site evaluation of fine-needle aspiration specimens: a review of 5,688 cases. Diagn Cytopathol, 27:1–4.

25.

Zhu

, Michael

. 2007. How important is on-site adequacy assessment for thyroid FNA? An evaluation of 883 cases. Diagn Cytopathol, 35:183–186.

26.

Sclabas

, Staerkel

, Shapiro

, Fornage

, Sherman

, Vassillopoulou-Sellin

, Lee

, Evans

. 2003. Fine-needle aspiration of the thyroid, correlation with histopathology in a contemporary series of 240 patients. Am J Surg, 186:702–709; discussion 09–10.

27.

Baloch

, Sack

, Yu

, LiVolsi

, Gupta

. 1998. Fine-needle aspiration of thyroid: an institutional experience. Thyroid, 8:565–569.

28.

Sakorafas

, Giotakis

, Stafyla

. 2005. Papillary thyroid microcarcinoma: a surgical perspective. Cancer Treat Rev, 31:423–438.

29.

Roti

, Rossi

, Trasforini

, Bertelli

, Ambrosio

, Busutti

, Pearce

, Braverman

, Degli Uberti

. 2006. Clinical and histological characteristics of papillary thyroid microcarcinoma: results of a retrospective study in 243 patients. J Clin Endocrinol Metab, 91:2171–2178.

30.

Kelman

, Rathan

, Leibowitz

, Burstein

, Haber

. 2001. Thyroid cytology and the risk of malignancy in thyroid nodules: importance of nuclear atypia in indeterminate specimens. Thyroid, 11:271–277.

31.

Raber

, Kaserer

, Niederle

, Vierhapper

. 2000. Risk factors for malignancy of thyroid nodules initially identified as follicular neoplasia by fine-needle aspiration: results of a prospective study of one hundred twenty patients. Thyroid, 10:709–712.

32.

Baloch

, Fleisher

, LiVolsi

, Gupta

. 2002. Diagnosis of “follicular neoplasm”: a gray zone in thyroid fine-needle aspiration cytology. Diagn Cytopathol, 26:41–44.

33.

, Yang

, Wasserman

, Bhuiya

, Griffith

, Michael

. 2006. Does Hürthle cell lesion/neoplasm predict malignancy more than follicular lesion/neoplasm on thyroid fine-needle aspiration? Diagn Cytopathol, 34:330–334.

34.

Poller

, Ibrahim

, Cummings

, Mikel

, Boote

, Perry

. 2000. Fine-needle aspiration of the thyroid. Cancer, 90:239–244.

35.

Khayyata

, Barroeta

, LiVolsi

, Baloch

. 2008. Papillary hyperplastic nodule: pitfall in the cytopathologic diagnosis of papillary thyroid carcinoma. Endocr Pract, 14:863–868.