Biopsy of Thyroid Nodules: Comparison of Three Sets of Guidelines

Neuroradiology/Head and Neck Imaging Original Research Ahn et al. Biopsy of Thyroid Nodules Neuroradiology/Head and Neck Imaging Original Research FOCUS ON: Sung Soo Ahn 1 Eun-Kyung Kim 1 Dae Ryong Kang 2 Sung-Kil Lim 3 Jin Young Kwak 1 Min Jung Kim 1 Ahn SS, Kim EK, Kang DR, Lim SK, Kwak JY, Kim MJ Keywords: fine-needle aspiration biopsy, thyroid, ultrasound DOI:10.2214/AJR.09.2822 Received March 30, 2009; accepted after revision July 7, 2009. 1 Department of Radiology and Research Institute of Radiological Science, Yonsei University, College of Medicine, 250 Seongsanno, Seodaemun-gu, Seoul 170-752, South Korea. Address correspondence to E. K. Kim (ekkim@yuhs.ac). 2 Clinical Trials Center, Yonsei University, College of Medicine, Seoul, South Korea. 3 Division of Endocrinology and Metabolism, Department of Internal Medicine, Yonsei University, College of Medicine, Seoul, South Korea. AJR 2010; 194:31 37 0361 803X/10/1941 31 American Roentgen Ray Society Biopsy of Thyroid Nodules: Comparison of Three Sets of Guidelines OBJECTIVE. The purpose of this study was to compare the results with three sets of guidelines for fine-needle aspiration biopsy of thyroid nodules. MATERIALS AND METHODS. A total of 1,398 nodules confirmed with fine-needle aspiration biopsy or surgery were included in the study. We compared the diagnostic value of three sets of guidelines for ultrasound findings that should lead to fine-needle aspiration biopsy of a nodule. According to the Kim criteria, a nodule should have at least one of the following findings: marked hypoechogenicity, irregular or microlobulated margins, microcalcifications, or length greater than width. According to the Society of Radiologists in Ultrasound, biopsy should be performed on a nodule 1 cm in diameter or larger with microcalcifications, 1.5 cm in diameter or larger that is solid or has coarse calcifications, and 2 cm in diameter or larger that has mixed solid and cystic components, and a nodule that has undergone substantial growth or is associated with abnormal cervical lymph nodes. According to the American Association of Clinical Endocrinologists, a hypoechoic nodule with at least one additional feature, such as irregular margins, length greater than width, and microcalcifications, should be biopsied. RESULTS. For all nodules, the diagnostic accuracy of the Kim (area under the receiver operating characteristic curve [A z ] = 0.868) and (A z = 0.850) criteria was greater than that of the Society of Radiologists in Ultrasound criteria (A z = 0.551). The number of nodules for which fine-needle aspiration biopsy was recommended (25.6%) was smallest with use of the American Association of Clinical Endocrinologists criteria, and the smallest number (7.3%) of missed malignant lesions was associated with use of the Kim criteria. The results did not change for the subgroup with nodules larger than 1 cm. CONCLUSION. The Kim and criteria are more accurate than the Society of Radiologists in Ultrasound criteria. The American Association of Clinical Endocrinologists guidelines are recommended for achieving high specificity, and the Kim criteria may be chosen for higher sensitivity. F ine-needle aspiration biopsy (FNAB) is considered the most reliable diagnostic test for evaluation of thyroid nodules and has a low rate of complications, especially when ultrasound guidance is used. Use of FNAB increases the rate of detection of thyroid cancer and reduces the number of unnecessary surgical procedures and overall cost of medical care [1 3]. However, performing FNAB on every thyroid nodule detected with ultrasound is not cost-effective because of the high prevalence of nodules. Studies have been conducted to identify sonographic features suggestive of malignancy. These findings include hypoechogenicity, irregular or microlobulated margins, calcifications, greater length than width, intranodular vascularity, solitary presence, and solid components [4 11]. No single ultrasound feature, however, has a high positive predictive value (PPV) for cancer that can be used to identify nodules that should be subjected to FNAB. For this reason, guidelines with various combinations of ultrasound features that are both sensitive and specific for predicting the presence of cancer have been defined in several studies, but debate continues about which guidelines are the most accurate and effective [4, 5, 7 9, 12, 13]. It is difficult to compare the accuracy of sets of guidelines, however, because of the differing samples in the studies in which the guidelines were established. In this study, we calculated the risk of malignancy of thyroid nodules with AJR:194, January 2010 31

Ahn et al. known suspicious sonographic features. We applied three sets of sonographic criteria from previous reports to the same sample to assess the diagnostic performance of the criteria in prediction of the presence of histologically confirmed malignancy. Materials and Methods The institutional review board of our university approved this study, and informed consent was not required. Informed consent for FNAB had been obtained from all patients before biopsy. (n = 128) Nondiagnostic (n = 161) (n = 33) (n = 32) Suspicious (n = 110) Thyroid Nodules Subjected to FNAB (n = 1,583) (n = 905) (n = 78) Cytology Benign (n = 1,016) (n = 111) Malignant (n = 244) (n = 38) Indeterminate (n = 52) (n = 25) (n = 206) Study Sample From September 2002 through July 2004, 1,583 nodules in 1,491 patients (1,371 women and girls [91.9%], 120 men [8%]; mean age, 49.0 years; range, 9 82 years) were biopsied under ultrasound guidance. Multiple nodules were present in 626 patients (42%). FNAB was performed on the largest of nodules with similar ultrasound features but on each nodule when multiple nodules had several different ultrasound features. One of five cytopathologists specializing in thyroid cytology interpreted the ultrasound-guided FNAB findings. The cytopathologist knew the ultrasound diagnosis made by the radiologist. At our institution, cytologic reports from ultrasound-guided FNAB indicate whether a specimen is adequate or inadequate. A specimen was considered adequate if a minimum of six groupings of well-preserved thyroid cells consisting of at least 10 cells per group was present [8]. The adequate specimens were further divided into the four subgroups benign, indeterminate, suspicious for papillary carcinoma, and malignancy. A benign cytologic finding included colloid nodules, nodular hyperplasia, lymphocytic thyroiditis, Graves disease, and postpartum thyroiditis. Indeterminate cytologic findings included follicular and Hürthle cell neoplasia. The finding suspicious for papillary carcinoma was used when the specimen exhibited cytologic atypia (nuclei crowded and overlapping, enlarged, and pleomorphic) but insufficient cellularity for a definite diagnosis of papillary carcinoma. Cytologic results were classified as malignancy when the specimen exhibited abundant cells with unequivocal cytologic features of cancer. Cytopathologic results obtained up to 2008 were reviewed retrospectively; the results of repeated biopsy were included in the data analysis. Among 1,583 nodules, thyroidectomy was performed for 455 nodules, including 33 with nondiagnostic cytologic findings, 111 benign nodules, 27 follicular neoplasms, 78 nodules suspicious for papillary carcinoma, and 206 malignant nodules (Fig. 1). Surgery was performed on nodules with benign cytologic findings for the following miscellaneous reasons: suspicious ultrasound findings in five patients, thyroid nodules showing substantial growth in 19, and compression symptoms or cosmetic problem due to large goiter (> 3 cm) in 40 patients. The other patients with thyroid nodules with benign cytologic findings underwent thyroid resection because they elected this method of treatment. Because surgery was not performed after FNAB, 128 of 161 nodules with nondiagnostic cytology, 25 of 52 nodules with cytologic findings of follicular neoplasm, and 32 of 110 nodules suspicious for papillary carcinoma were excluded from this study. Nodules with benign (n = 1,016) or malignant (n = 244) cytologic findings were included whether or not the patient underwent surgery. Sonographic Evaluation and Ultrasound- Guided FNAB Thyroid ultrasound (HDI 3000 or 5000 unit, Philips Healthcare) was performed with a 7- to 12- MHz transducer. All images were sent to the local PACS for review. One experienced radiologist who performed FNAB prospectively described the sonographic characteristics of thyroid nodules with respect to the following parameters: size, multiplicity, composition, echogenicity, margin, calcification, shape, and abnormal cervical lymph nodes. Size was measured at the maximum dimension. Substantial growth was retrospectively assessed in 287 nodules examined with ultrasound at least 6 months before FNAB. The composition of the nodule was classified as purely cystic, mixed, or solid on the basis of the ratio of the cystic portion to the solid portion in the nodule. A purely cystic nodule was completely anechoic with or without a comet-tail artifact. A nodule having both a cystic and solid portion was defined as a mixed echoic nodule. Echogenicity of a mixed echoic nodule was assessed on the basis of the solid portion. The echogenicity of the nodule was compared with that of the surrounding parenchyma and was classified as hypoechogenic, isoechogenic, or hyperechogenic. Marked hypoechogenicity was defined as lower echogenicity than the cervical strap muscle. (n = 27) Study Population (n = 1,398) Benign (n = 1,071) Malignant (n = 327) Fig. 1 Study sample. FNAB = fine-needle aspiration biopsy. 32 AJR:194, January 2010

Biopsy of Thyroid Nodules The margin of a nodule was described as well circumscribed or not well circumscribed, which included irregular or microlobulated margins. Microcalcifications were defined as tiny (< 2 mm diameter), punctuate, and hyperechoic foci with or without acoustic shadows. Macrocalcifications or coarse calcifications were defined as larger than 2 mm. Mixed calcifications were defined as a combination of microcalcifications and macrocalcifications, which were reclassified as the presence of microcalcifications. Shape was assessed as the ratio of anteroposterior to transverse diameter ( 1 or < 1). Abnormal lymph nodes were defined as lymph nodes with heterogeneous echotexture, calcifications, cystic areas, and round shape. Ultrasound-guided FNAB was performed by one experienced radiologist using a 23-gauge needle attached to a 20-mL disposable plastic syringe and aspirator. Each lesion was aspirated at least twice. Materials obtained at aspiration biopsy were expelled onto glass slides and smeared. All smears were placed immediately in 95% alcohol for Papanicolaou staining. The rest of the material was rinsed in saline solution for processing as a cell block. The cytopathologist was not on site during the biopsy. Additional special staining was made on a case-by-case basis according to the requirements of the cytopathologist. Comparison of Guidelines Three sonographic criteria were applied to our data to compare the accuracies of the various guidelines in prediction of cancer. In addition, we regrouped the study sample with nodules 1 cm or more in diameter and analyzed them in the same way. Kim criteria According to the Kim criteria [4], FNAB of nodules with any single suspicious Fig. 2 42-year-old woman with thyroid nodule. Transverse ultrasound image of left lobe of thyroid shows 10-mm wider than long (anteroposterior diameter, 7 mm; transverse diameter, 10 mm) hypoechoic nodule with irregular margin (arrow). These ultrasound findings would lead to fine-needle aspiration biopsy on basis of Kim and American Association of Clinical Endocrinologists criteria. Nodule was confirmed papillary carcinoma with extrathyroidal invasion. Fig. 3 47-year-old woman with thyroid nodule. Transverse ultrasound image of thyroid shows 7-mm well-defined, longer than wide (anteroposterior diameter, 7 mm; transverse diameter, 4 mm) isoechoic nodule (arrow). Fine-needle aspiration biopsy was recommended only on basis of Kim criteria, and nodule was confirmed papillary carcinoma with extrathyroidal invasion. ultrasound feature, regardless of nodule size, is recommended. Suspicious sonographic features are defined as marked hypoechogenicity, irregular or microlobulated margins, microcalcifications, and anteroposterior to transverse diameter ratio of 1 or greater (Fig. 2). Society of Radiologists in Ultrasound criteria The Society of Radiologists in Ultrasound criteria [7] are based on the size and ultrasound characteristics of thyroid nodules. The recommendations for FNAB are as follows: nodule 1.0 cm or more in diameter if microcalcifications are present; nodule 1.5 cm or more in diameter if completely or almost entirely solid or if coarse calcifications (classified as macrocalcifications) are present; nodule 2.0 cm or more in diameter if mixed solid and cystic components are present or nodule is almost entirely cystic with a solid mural component; and nodule has grown substantially since previous ultrasound examination (Figs. 3 and 4). Because there was no consensus on the definition of substantial growth that would necessitate biopsy, we included nodules with an increase in maximal diameter of 3 mm or more [14]. This guideline also calls for FNAB when abnormal lymph nodes are present regardless of the ultrasound features of thyroid nodules. Ultrasound features associated with high risk of cancer are defined as lymph nodes with heterogeneous echotexture, calcifications, cystic areas, round shape, and mass effect. Because lymph node size is not mentioned, we included morphologically abnormal lymph nodes regardless of size. criteria According to the American Association of Clinical Endocrinologists guidelines [8], FNAB should be performed on all hypoechoic nodules with at least one of the following additional ultrasound features: irregular margins, intranodular vascular spots, longer than wide shape, or microcalcifications (Fig. 2). The parameter intranodular vascularity was excluded from our study because color Doppler sonography was not performed. Statistical Analyses Statistical analysis was performed with statistical software (SAS version 9.1.3, SAS; MedCalc version 9.3.6.0, MedCalc). The odds ratio of malignancy for each ultrasound feature was calculated with logistic regression analysis. We calculated sensitivity, specificity, PPV, negative predictive value (NPV), and accuracy for individual sonographic characteristics and various combinations of sonographic features. The diagnostic accuracy of predictions of malignancy was calculated with receiver operating characteristic analysis of each set of guidelines. The number of nodules recommended for FNAB and the rate of missed carcinoma were assessed on the basis of the criteria, and statistical significances was evaluated with Yates correction for continuity. Results A total of 1,398 nodules in 1,318 patients (1,217 women, 101 men; mean age, 46.3 years) were included in this study. The final diagnosis was based on surgical pathologic findings or on cytologic findings if the patient did not undergo surgery. There were 1,071 benign (76.6%) and 327 malignant (23.4%) nodules (Table 1). The diagnosis of malignancy included papillary carcinoma (n = 319), follicular carcinoma (n = 3), lymphoma (n = 1), medullary carcinoma (n = 1), anaplastic carcinoma (n = 1), metastasis from breast cancer (n = 1), and poor- Fig. 4 32-year-old woman with thyroid nodule. Transverse ultrasound image of thyroid shows 4-cm well-defined mixed solid and cystic nodule. Fineneedle aspiration biopsy would be recommended only with Society of Radiologists in Ultrasound criteria. Surgical finding was adenomatous hyperplasia. AJR:194, January 2010 33

Ahn et al. TABLE 1: Size and Pathologic Features of Thyroid Nodules in Study Sample (n = 1,398) Size Benign Malignant Total < 1 cm 233 (58.0) 169 (42.0) 402 1 cm 838 (84.1) 158 (15.9) 996 Total 1,071 (76.6) 327 (23.4) 1,398 Note Values in parentheses are percentages. ly differentiated carcinoma (n = 1). The mean size of the nodules was 16.6 mm (range, 3 100 mm); 402 nodules (28.8%) were smaller than 1 cm in diameter. The mean size of benign nodules was 17.9 mm (range, 3 100 mm) and of malignant nodules was 12.1 mm (range, 3 80 mm). Among 287 nodules that had been examined with ultrasound at least 6 months before FNAB, 36 nodules (12.5%) had an increase in maximal diameter of 3 mm or more. Ultrasound Findings Independent predictors of the presence of malignant thyroid nodules at ultrasound examination are irregular or microlobulated margins, microcalcifications, marked hypoechogenicity, hypoechogenicity, longer than wide shape, solidity, and presence of calcifications (Table 2). These seven ultrasound findings had significant associations with the presence of malignancy (p < 0.001). Morphologically abnormal lymph nodes were found only in cases of malignancy (17/327). Solitary or multiple nature was not associated with risk of malignancy (odds ratio, 1.1). The frequency of each suspicious ultrasound feature based on nodule size is shown in Table 3. All suspicious findings, except abnormal lymph node, were made significantly more frequently in nodules smaller than 1 cm. Diagnostic Performance According to the Three Sets of Guidelines The sensitivity, specificity, PPV, and NPV of the Kim criteria were estimated to be 92.7%, 80.9%, 59.6%, and 97.3%. The area under the receiver operating characteristic curve (A z ) based on the Kim criteria was 0.868. The A z value based on the Society of Radiologists in Ultrasound criteria was 0.551 (p < 0.001), and the A z based on the American Association of Clinical Endocrinologists criteria was 0.850 (p = 0.179) (Fig. 5). The sensitivity, specificity, PPV, and NPV of each set of criteria are shown in Table 4. In a comparison of the A z of the Kim criteria with that of the other criteria for nodules 1 cm or more in diameter (Table 4), the A z of the Kim criteria was 0.872, significantly greater than that of the Society of Radiologists in Ultrasound criteria (0.567; p < 0.001). The Kim criteria appeared to be more accurate than the American Association of Clinical Endocrinologists criteria for nodules 1 cm in diameter or larger (A z = 0.872 vs 0.842; p = 0.117), but the difference was not significant. The three sets of guidelines were compared to determine the number of biopsies that would have been performed and the number of cases of cancer that would have been missed had the criteria been applied (Table 5). For all nodules, the percentage of nodules recommended for FNAB on the basis of the criteria could have been reduced to 25.6% (358/1,398), significantly lower than that based on the other criteria (p < 0.001). TABLE 2: Diagnostic Accuracy of Ultrasound Findings for Presence of Malignant Nodules However, the percentage of malignant tumors missed with use of the Kim criteria (7.3%) was the lowest (p < 0.001). When we applied the criteria to thyroid nodules 1 cm in diameter or larger, use of the American Association of Clinical Endocrinologists criteria would have decreased the number of nodules for which FNAB is recommended to 16.5%, and the Kim criteria showed low risk of missing carcinoma (10.8%). Discussion Detection of thyroid nodules has increased with the wide use of ultrasound, and it is important to exclude the presence of malignancy, which is found in approximately 5% of all thyroid nodules regardless of size [8]. Because of the high prevalence of thyroid nodules, it is essential to determine a reliable Characteristic Odds Ratio Sensitivity Specificity Positive Predictive Value Negative Predictive Value A z Irregular or microlobulated margin 30.811 276/327 (84.4) 911/1,071 (85.1) 276/436 (63.3) 911/962 (94.7) 0.847 Microcalcifications 28.985 120/327 (36.7) 1,050/1,071 (98.0) 120/141 (85.1) 1,050/1,257 (83.5) 0.674 Marked hypoechogenicity 19.448 120/327 (36.7) 1,040/1,071 (97.1) 120/151 (79.5) 1,040/1,247 (83.4) 0.669 Hypoechogenicity 18.387 265/327 (81.0) 869/1,071 (81.1) 265/467 (56.7) 869/931 (93.9) 0.811 Longer than wide 17.633 183/327 (56.0) 999/1,071 (93.3) 183/255 (71.8) 999/1,143 (87.4) 0.746 Solid 11.338 292/327 (89.3) 617/1,071 (57.6) 292/746 (39.1) 617/652 (94.6) 0.735 Calcifications 9.836 149/327 (45.6) 987/1,071 (92.2) 149/233 (63.9) 987/1,165 (84.7) 0.689 Solitary 1.053 187/327 (57.2) 472/1,071 (44.1) 187/786 (23.8) 472/612 (77.1) 0.506 Diameter 1 cm 0.260 158/327 (48.3) 233/1,071 (21.8) 158/996 (15.9) 233/402 (58.0) 0.65 Abnormal lymph nodes Not calculable 17/327 (5.2) 1,071/1,071 (100) 17/17 (100) 1,071/1,381 (77.6) 0.526 Substantial growth Not calculable Not calculable Not calculable 2/36 (5.6) 193/251 (76.9) Not calculable Note Values in parentheses are percentages. A z = area under receiver operating characteristic curve. Sensitivity (%) 100 80 60 40 20 0 0 20 40 60 80 100 100 Specificity (%) Fig. 5 Graph shows receiver operator characteristic curves for Kim criteria (solid curve), Society of Radiologists in Ultrasound criteria (dashed curve), and American Association of Clinical Endocrinologists criteria (dotted curve). 34 AJR:194, January 2010

Biopsy of Thyroid Nodules TABLE 3: Frequency of Suspicious Ultrasound Features According to Nodule Size Characteristic Diameter < 1 cm (n = 402) Diameter 1 cm (n = 996) p Irregular or microlobulated margin 215 (53.5) 221 (22.2) < 0.001 Microcalcifications 52 (12.9) 89 (8.9) 0.025 Marked hypoechogenicity 97 (24.1) 54 (5.4) < 0.001 Hypoechogenicity 226 (56.2) 241 (24.2) < 0.001 Longer than wide 159 (39.6) 96 (9.6) < 0.001 Solid 293 (72.9) 453 (45.5) < 0.001 Calcifications 83 (20.6) 150 (15.1) 0.014 Solitary 227 (56.5) 559 (56.1) 0.954 Abnormal lymph nodes 3 (0.7) 14 (1.4) 0.309 Note Values are numbers of nodules with percentages in parentheses. TABLE 4: Diagnostic Accuracy of Each Set of Criteria All nodules and cost-effective guideline for basing the management of thyroid nodules on the ultrasound features. In a multicenter retrospective study, spiculated margins, microcalcifications, marked hypoechogenicity, longer than wide shape, and macrocalcifications had a significant association with the presence of malignancy [15]. These suspicious features had relatively high specificity (80.8 96.1%) and PPV (64.8 81.3%) even though the sensitivity was low (9.7 48.3%). No single sonographic feature, however, is accurate enough for a recommendation of FNAB; therefore, several guidelines with various combinations of suspicious ultrasound findings have been suggested [4, 7, 8]. In this study, the A z of the Kim and American Association of Clinical Endocrinologists criteria were comparable (0.868 and 0.850). Both of these values were superior to that of the Society of Radiologists in Ultrasound criteria (A z = 0.551). The Kim criteria were originally suggested for recommendations of FNAB of nonpalpable solid nodules of the thyroid and had high sensitivity (93.8%) and NPV (95.9%). Tae et al. [16] applied the same FNAB criteria to palpable or nonpalpable thyroid nodules and also found a high NPV (97.8%) and diagnostic accuracy (86.5%). These results suggest that the Kim criteria can be applied not only to nonpalpable nodules but also palpable nodules. The Society of Radiologists in Ultrasound recommendations are based on the size and ultrasound characteristics of thyroid nodules measuring 1 cm or more in largest diameter. They do not call for FNAB of a nodule smaller than 1 cm because it is assumed that thyroid surgery for microcarcinoma may not improve life expectancy owing to the indolent clinical course of this tumor. However, the prevalence of thyroid cancer does not differ between nodules larger or smaller than 1 cm [5, 6]. Furthermore, microcarcinoma can have an aggressive course, including extracapsular growth and nodal metastasis [5, 6, 17 21]. The overall malignancy rate in our study was 23.4% (327/1,398). This high rate was due to the high prevalence of cancer in nodules smaller than 1 cm (169/402, 42%). The Criteria Sensitivity (%) Specificity (%) Positive Predictive Value (%) Negative Predictive Value (%) A z p a Kim 92.7 80.9 59.6 97.3 0.868 Society of Radiologists in Ultrasound 35.5 54.3 19.2 80.8 0.551 < 0.001 American Association of Clinical Endocrinologists Nodules 1 cm in diameter 79.2 90.8 72.3 93.5 0.850 0.179 Kim 89.2 85.2 53.2 97.7 0.872 Society of Radiologists in Ultrasound 71.5 41.9 18.8 88.6 0.567 < 0.001 American Association of Clinical Endocrinologists Note A z = area under receiver operating characteristic curve. a From receiver operating characteristic comparison with Kim criteria. 74.1 94.4 71.3 95.1 0.842 0.117 TABLE 5: Biopsies That Would Have Been Performed and Cases of Thyroid Carcinoma That Would Have Been Missed With Each Set of Criteria Criteria All Nodules (n = 1,398) Nodules 1 cm (n = 996) Nodules Recommended for FNAB a Missed Carcinoma (n = 327) Nodules Recommended for FNAB a Missed Carcinoma (n = 158) Kim 508 (36.3) 24 (7.3) 265 (26.6) 17 (10.8) Society of Radiologists in Ultrasound American Association of Clinical Endocrinologists 605 (43.8) 211 (64.5) 600 (60.2) 45 (28.5) 358 (25.6) 68 (20.8) 164 (16.5) 41 (25.9) Note Values are numbers of nodules with percentage in parentheses. FNAB = fine-needle aspiration biopsy. a Number of biopsies that would have been performed had the criteria been applied to the overall sample of nodules. AJR:194, January 2010 35

Ahn et al. malignancy rate also may have been high because many patients with small suspicious thyroid nodules had been referred to our tertiary care hospital for FNAB. In other words, patients with small nodules with obviously benign ultrasound findings might not have been referred for FNAB. Other authors [6, 22] also have reported a high prevalence of malignancy of subcentimeter nodules compared with larger nodules. In this series, suspicious findings were more prevalent in nodules smaller than 1 cm than in nodules larger than 1 cm (Table 3). This observation suggests that the size of nodules may be a poor predictor of malignancy. The Society of Radiologists in Ultrasound suggests that FNAB be considered for nodules exhibiting substantial growth on serial ultrasound images. However, only two of 36 nodules that grew more than 3 mm proved to be malignant and consequently had a low PPV (5.6%). Our findings support previous results suggesting that the presence or absence of growth is not a reliable marker of the malignant or benign nature of a nodule [8, 9, 23 26]. According to the American Association of Clinical Endocrinologists criteria, additional suspicious ultrasound features include intranodular vascular spots. We did not use this criterion because our data did not include color Doppler findings. Color Doppler ultrasound is useful for evaluating intranodular and perinodular vascularity in the thyroid, but its utility for differentiating benign from malignant nodules continues to be debated [5, 27 30]. Results of several previous studies [5, 28, 29] have suggested that intranodular vascularity is a significant indicator of malignancy and that the sensitivity of intranodular vascularity is relatively high, ranging from 66.7% to 91.7%. In those studies, however, specificity varied (range, 34.7 80.8%) and PPV was low (range, 23.2 34.5%), possibly because intranodular vascularity was identified in many benign nodules. If we had included color Doppler sonographic data, the false-positive rate likely would have been higher and the specificity and PPV of the criteria lower, but sensitivity would have increased. Further study that includes data on the type of vascularity is needed for direct comparison of the diagnostic accuracies of the various criteria. The American Thyroid Association (ATA) recommends routine FNAB of nodules 1 cm in diameter or larger unless the serum level of thyroid-stimulating hormone is low [9]. According to the ATA, nodules smaller than 1 cm may have to be evaluated if the ultrasound findings are suspicious. However, because the suspicious ultrasound findings were not specified [9], we did not include the ATA guideline in this analysis. If we had performed FNAB on nodules measuring 1 cm or larger, according to the ATA guidelines, 996 nodules would have been biopsied to identify 158 malignant lesions. Several guidelines, including those of the Society of Radiologists in Ultrasound, the, and the ATA, and routine observation have been evaluated in terms of relative desirability [31]. The investigators reported that use of all of the guidelines except those of the ATA had similar outcomes and that performance of FNAB on the basis of the ATA guidelines appeared to be the least desirable course. The baseline model assumptions in that study, however, were based on the likelihood suggested in previous studies, and the investigators calculated the probability of every potential outcome statistically instead of estimating it in real patients. Therefore, the findings may not necessarily be applied to the general population. In contrast, we applied ultrasound criteria to a large set of real nodules in patients. Our study had limitations. First, although many patients had multiple nodules, we did not perform FNAB on all nodules, which might have caused selection bias. However, because we do not perform FNAB on all thyroid nodules detected with ultrasound in clinical practice, this limitation may be inevitable. Second, all sonographic evaluations were performed by one experienced radiologist, so we did not include interobserver variation data. Moon et al. [15] reported that reviewers had fair to high agreement on the ultrasound characteristics of thyroid nodules (κ > 0.5). A large-scale prospective study would be warranted to confirm these results. Third, we did not include color Doppler findings as part of the sonographic evaluation. Last, we did not have clinical information, such as previous head and neck irradiation, family history of thyroid cancer, or serum thyroid-stimulating hormone or thyroglobulin level, that might have affected the decision to perform FNAB. This lack of information, however, might not necessarily have affected the results because all of the criteria were applied to the same nodules under the same conditions with blinding to clinical history. We conclude that the Kim and American Association of Clinical Endocrinologists criteria are more accurate than the Society of Radiologists in Ultrasound criteria. The guidelines are recommended for achieving high specificity and the Kim criteria for achieving high sensitivity. References 1. Hamberger B, Gharib H, Melton LJ, Goellner JR, Zinsmeister AR. Fine-needle aspiration biopsy of thyroid nodules: impact on thyroid practice and cost of care. Am J Med 1982; 73:381 384 2. Mittendorf EA, Tamarkin SW, McHenry CR. The results of ultrasound-guided fine-needle aspiration biopsy for evaluation of nodular thyroid disease. Surgery 2002; 132:648 653; discussion 653 3. Hegedüs L. Clinical practice: the thyroid nodule. N Engl J Med 2004; 351:1764 1771 4. Kim EK, Park CS, Chung WY, et al. New sonographic criteria for recommending fine-needle aspiration biopsy of nonpalpable solid nodules of the thyroid. AJR 2002; 178:687 691 5. Papini E, Guglielmi R, Bianchini A, et al. Risk of malignancy in nonpalpable thyroid nodules: predictive value of ultrasound and color-doppler features. J Clin Endocrinol Metab 2002; 87:1941 1946 6. Nam-Goong IS, Kim HY, Gong G, et al. Ultrasonography-guided fine-needle aspiration of thyroid incidentaloma: correlation with pathological findings. Clin Endocrinol (Oxf) 2004; 60:21 28 7. Frates MC, Benson CB, Charboneau JW, et al. Management of thyroid nodules detected at US: Society of Radiologists in Ultrasound consensus conference statement. Radiology 2005; 237:794 800 8. Gharib H, Papini E, Valcavi R, et al. American Association of Clinical Endocrinologists and Associazione Medici Endocrinologi medical guidelines for clinical practice for the diagnosis and management of thyroid nodules. Endocr Pract 2006; 12:63 102 9. Cooper DS, Doherty GM, Haugen BR, et al. Management guidelines for patients with thyroid nodules and differentiated thyroid cancer. Thyroid 2006; 16:109 142 10. Chan BK, Desser TS, McDougall IR, Weigel RJ, Jeffrey RB. Common and uncommon sonographic features of papillary thyroid carcinoma. J Ultrasound Med 2003; 22:1083 1090 11. Peccin S, de Castsro JA, Furlanetto TW, Furtado AP, Brasil BA, Czepielewski MA. Ultrasonography: is it useful in the diagnosis of cancer in thyroid nodules? J Endocrinol Invest 2002; 25:39 43 12. Butros R, Boyvat F, Ozyer U, et al. Management of infracentimetric thyroid nodules with respect to ultrasonographic features. Eur Radiol 2007; 17:1358 1364 36 AJR:194, January 2010

Biopsy of Thyroid Nodules 13. Cappelli C, Castellano M, Pirola I, et al. The pre- metastasis from 259 papillary thyroid microcarci- Thyroid 2004; 14:953 958 dictive value of ultrasound findings in the manage- nomas: frequency, pattern of occurrence and re- 26. Roti E. degli Uberti EC, Bondanelli M, Braver- ment of thyroid nodules. QJM 2007; 100:29 35 currence, and optimal strategy for neck dissec- man LE. Thyroid papillary microcarcinoma: a 14. Alexander EK, Hurwitz S, Heering JP, et al. Natu- tion. Ann Surg 2003; 237:399 407 descriptive and meta-analysis study. Eur J Endo- ral history of benign solid and cystic thyroid nod- 21. Cappelli C, Castellano M, Braga M, et al. Aggres- crinol 2008; 159:659 673 ules. Ann Intern Med 2003; 138:315 318 siveness and outcome of papillary thyroid carci- 27. Shimamoto K, Endo T, Ishigaki T, Sakuma S, 15. Moon WJ, Jung SL, Lee JH, et al. Benign and malignant thyroid nodules: US differentiation multicenter retrospective study. Radiology 2008; 247: 762 770 16. Tae HJ, Lim DJ, Baek KH, et al. Diagnostic value of ultrasonography to distinguish between benign and malignant lesions in the management of thyroid nodules. Thyroid 2007; 17:461 466 17. Mazzaferri EL, de los Santos ET, Rofagha-Keyhani S. Solitary thyroid nodule: diagnosis and management. Med Clin North Am 1988; 72:1177 1211 18. Noguchi S, Yamashita H, Murakami N, Nakayama I, Toda M, Kawamoto H. Small carcinomas of the thyroid: a long-term follow-up of 867 patients. Arch Surg 1996; 131:187 191 19. Berker D, Aydin Y, Ustun I, et al. The value of fine-needle aspiration biopsy in subcentimeter thyroid nodules. Thyroid 2008; 18:603 608 20. Wada N, Duh QY, Sugino K, et al. Lymph node noma (PTC) versus microcarcinoma (PMC): a mono-institutional experience. J Surg Oncol 2007; 95:555 560 22. Sahin M, Sengul A, Berki Z, Tutuncu NB, Guvener ND. Ultrasound-guided fine-needle aspiration biopsy and ultrasonographic features of infracentimetric nodules in patients with nodular goiter: correlation with pathological findings. Endocr Pathol 2006; 17:67 74 23. Filetti S, Durante C, Torlontano M. Nonsurgical approaches to the management of thyroid nodules. Nat Clin Pract Endocrinol Metab 2006; 2:384 394 24. Quadbeck B, Pruellage J, Roggenbuck U, et al. Long-term follow-up of thyroid nodule growth. Exp Clin Endocrinol Diabetes 2002; 110:348 354 25. Alexander EK, Marqusee E, Orcutt J, et al. Thyroid nodule shape and prediction of malignancy. Makino N. Thyroid nodules: evaluation with color Doppler ultrasonography. J Ultrasound Med 1993; 12:673 678 28. Rago T, Vitti P, Chiovato L, et al. Role of conventional ultrasonography and color flow-doppler sonography in predicting malignancy in cold thyroid nodules. Eur J Endocrinol 1998; 138:41 46 29. Appetecchia M, Solivetti FM. The association of colour flow Doppler sonography and conventional ultrasonography improves the diagnosis of thyroid carcinoma. Horm Res 2006; 66:249 256 30. Frates MC, Benson CB, Doubilet PM, Cibas ES, Marqusee E. Can color Doppler sonography aid in the prediction of malignancy of thyroid nodules? J Ultrasound Med 2003; 22:127 131; quiz 132 31. McCartney CR, Stukenborg GJ. Decision analysis of discordant thyroid nodule biopsy guideline criteria. J Clin Endocrinol Metab 2008; 93:3037 3044 AJR:194, January 2010 37