World J Surg (2008) 32:1948–1954 DOI 10.1007/s00268-008-9600-2
Predictive Value of Sonographic Features in Preoperative Evaluation of Malignant Thyroid Nodules in a Multinodular Goiter _ ¸ sever Æ Artu¨r Salmaslıog˘lu Æ Yes¸ im Erbil Æ Cem Dural Æ Halim Is ¨ Yersu Kapran Æ Selc¸uk Ozarmag˘an Æ Serdar Tezelman
Published online: 29 April 2008 Ó Socie´te´ Internationale de Chirurgie 2008
Abstract Background The primary goal of ultrasonography (US) in the evaluation of a thyroid nodule is to determine its malignancy, although the diagnosis of a malignant nodule on the basis of US alone is nearly impossible. The aim of this prospective study was to evaluate the predictive value of sonographic features in the preoperative diagnosis of malignant thyroid nodules, and to determine the important features of sonography. Methods This prospective study included 550 consecutive patients with , thyroid nodules. Nodules were divided into two groups on the basis of pathological diagnosis: group 1 consisted of 1,633 nodules with a benign pathology, and group 2 consisted of 293 nodules with a malignant pathology. Results Microcalcifications, blurred nodular margins, and solid and hypoechoic appearance were more common in malignant nodules compared to benign nodules (89.1% versus 5%; 64.5% versus 4.7%; 81.6% versus 30.6% ; and 62.5% versus 43.1%, respectively; p \ 0.001). There was a
A. Salmaslıog˘lu Department of Radiology, Istanbul Medical Faculty, Istanbul University, Capa, 34093 Istanbul, Turkey ¨ zarmag˘an S. Tezelman Y. Erbil (&) C. Dural S. O Department of General Surgery, Istanbul Medical Faculty, Istanbul University, Capa, 34093 Istanbul, Turkey e-mail:
[email protected] _ ¸ sever H. Is Department of Public Health, Istanbul Medical Faculty, Istanbul University, Capa, 34093 Istanbul, Turkey Y. Kapran Department of Pathology, Istanbul Medical Faculty, Istanbul University, Capa, 34093 Istanbul, Turkey
123
positive correlation between the detection of malignant thyroid nodules and microcalcification (rs = 0.791, p = 0.0001), blurred nodular margin (rs = 0.625, p = 0.0001), solid appearance (rs = 0.376, p = 0.0001), and hypoechoic appearance (rs = 0.141, p = 0.0001). Microcalcifications, blurred nodular margins, and solid and hypoechoic appearance were independent determinants of malignancy upon US examination of thyroid nodules (OR: 159, OR: 37, OR: 9.9, and OR: 2.2, respectively). Conclusion Although we did not identify a single feature indicative of malignancy in the sonographic examination of nodules, microcalcification and blurred margin were the strongest correlates for malignancy.
Introduction Thyroid nodules are very common in the population, with an estimated prevalence ranging from 4% detected through palpation to 67% identified through ultrasonography (US) [1, 2]. The majority of thyroid nodules are diagnosed incidentally during physical examination or radiographic studies [3]. Generally, thyroid nodules are benign and require no surgical intervention. Thus, the distinction between malignant and benign thyroid nodules is a key determinant of treatment [2, 4]. Fine-needle aspiration biopsy (FNAB) has become one of the most useful, safe, and accurate tools in the diagnosis of thyroid nodules [5]. The decision to perform a biopsy is based on multiple factors including patient-specific factors and US findings [5–7]. A history of neck irradiation, rapid tumor growth, or a family history increases the risk of thyroid carcinoma. Physical examination factors associated with increased likelihood of malignancy include firmness of the nodule, rapid growth, fixation to adjacent structures,
World J Surg (2008) 32:1948–1954
vocal cord paralysis in the absence of previous surgery, and enlarged regional lymph nodes [2–4]. Ultrasonography has emerged as a widely used tool in the evaluation of thyroid nodules [3, 6–8], largely because of its availability, low cost, limited discomfort to the patient, and non-ionizing nature. The advent of high-resolution US has led to the discovery of small, asymptomatic, and previously unrecognized thyroid nodules [6, 7]. These thyroid incidentalomas are often diagnosed during US evaluation for nonthyroid neck disease, posing a management dilemma for the clinician [4, 5, 9]. The optimal diagnostic strategy for a patient with multinodular goiter is still the subject of debate. Fine-needle aspiration from all palpable or nonpalpable nodules is impractical, and therefore the clinician must select specific nodules for biopsy. Several US features have been found to be associated with an increased risk of thyroid carcinoma: the presence of calcifications, hypoechogenicity, irregular margins, absence of a halo, predominantly solid composition, and intranodule vascularity [10–16]. Although US has emerged as a widely used tool in the evaluation of thyroid nodules, the overall sensitivity is extremely variable from study to study. The aims of this prospective clinical study were (1) to evaluate the predictive value of sonographic features in the preoperative diagnosis of malignant thyroid nodules and (2) to determine which sonographic features are important.
Materials and methods Patients From January 2006 to June 2007, 610 patients with multinodular goiter were referred for thyroidectomy to the Department of General Surgery, Istanbul Medical Faculty. The patients lived in an iodine-deficient area. This prospective study included 550 consecutive patients with 1,926 thyroid nodules larger than 3 mm in diameter. The indications for surgical treatment were as follows: large goiter (n = 290), toxic multinodular goiter (n = 75), and findings suggestive of malignancy in an FNAB (n = 185). Patients with thyroid malignancies other than papillary carcinoma or patients refusing to participate in this study were excluded. Total and near-total thyroidectomies were performed in all patients. The final pathological diagnosis was determined by a single pathologist (Y.K.) experienced in endocrine pathology. The pathologist was blinded to the US features. Nodules were divided into two groups on the basis of pathological diagnosis: group 1 consisted of 1,633 (84.7%) nodules with benign pathology, and group 2 consisted of 293 (15.3%) nodules with a malignant pathology (papillary
1949
carcinoma). The study plan was reviewed and approved by our institutional ethical committee, and informed consent was obtained from all patients. Ultrasonography Ultrasound examinations were performed with a digital US scanner equipped with an 11.4-MHz linear transducer (Sonoline Antares, Siemens, Erlangen, Germany) for morphological studies. Examinations were conducted and recorded by a skilled sonographer (A.S.) according to a standard procedure. The following ultrasound parameters were assessed in all nodules: (1) nodule size (maximum diameter as evaluated by sagittal and transverse scans); (2) echographic structure (solid, mixed or cystic); (3) echogenicity (iso-, hyper- or hypoechoic); (4) presence/absence of calcification (microcalcification, coarse calcification, peripheric calcification); (5) lesion margins well-defined or blurred. Preoperatively identified calcified, ill-defined, solid, hypoechoic nodules were considered to be malignant. The strap muscle is uniformly present in all patients; therefore, we chose the strap muscle as the comparative standard for the evaluation of the echogenicity of solid nodules. Statistical analysis Clinical, ultrasound, cytological, and histological findings were separately recorded and processed for blinded statistical evaluation. Analysis was performed with the statistical package SPSS 10.1 (SPSS, Chicago, IL). Differences between parameters were compared with Student’s t-test. Frequency distributions were compared with the chi-square test. Univariate odds ratios (OR) with 95% confidence intervals (CI) were calculated to assess the relationships between ultrasound criteria and histological outcomes. Results were considered statistically significant at p \ 0.05. The specific ultrasonographic criteria and final histopathological results of nodules were analyzed and compared. The numbers of true-positive (TP), true-negative (TN), false-positive (FP), and false-negative (FN) US diagnoses were determined. The diagnostic value of US criteria was also assessed in terms of sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), diagnostic accuracy, and conditional probability.
Results Patients The mean age of the patients was 46.9 ± 12 (16–82) years, with a female/male ratio (F/M) of 4.2/1 (n = 445/105).
123
1950
World J Surg (2008) 32:1948–1954
Histopathological examination revealed papillary thyroid carcinoma in 211 patients (38%) and benign thyroid pathology in 339 patients (62%). The histopathological criteria of thyroid carcinomas included the presence of thyroid capsule invasion in 118 patients (56%) and vascular invasion in 37 patients (16%). The carcinomas were multifocal in 35 (16.5%) patients (two foci in 23 patients and three foci in 12 patients). Histopathological evaluation revealed a benign pathology in 1,633 (76.3%) nodules and a malignant pathology in 293 (5.7%) nodules. The mean nodule size was 19.3 ± 13 mm (range: 3–89 mm). Overall, 1,926 nodules were examined, of which 535 (27.8%) were \1 cm in diameter.
The PPV of microcalcification and blurred nodular margins was greater than solid and hypoechoic appearance (76% and 71.3% versus 32.2% and 20.6%, respectively; p \ 0.001) (Figs. 3 and 4). There was no significant difference in NPV between the microcalcification, blurred nodular margins and solid appearance. Diagnostic accuracy of microcalcification and blurred nodular margin (94.1% and 90.7%, respectively) in preoperative diagnosis of thyroid carcinoma was significantly greater compared to other ultrasonographic features (p \ 0.001). The predictive values of ultrasonographic features are summarized in Table 2. Correlations
Histopathologic findings correlated with sonographic features There were no significant differences in nodule size between malignant and benign nodules (19.8 ± 15 mm versus 19.2 ± 13 mm, respectively; p [ 0.05). Microcalcifications, blurred nodular margins, and solid and hypoechoic appearance were more common in malignant nodules than in benign nodules (89.1% versus 5%; 64.5% versus 4.7%; 81.6% versus 30.6%; and 62.5% versus 43.1%, repectively; p \ 0.001). Benign nodules were generally well defined (95.3%) in this study. Hyperechogenicity was relatively rare within both groups. The relationships between ultrasound findings and histological features are shown in Table 1.
The detection of malignant thyroid nodules by US did not correlate with patient age (rs = -0.029, p = 0.199), gender (rs = 0.011, p = 0.633), or the nodule size (rs = 0.16, p = 0.482). However, there was a positive correlation between the detection of malignant thyroid nodules and microcalcification (rs = 0.791, p = 0.0001), blurred nodular margin (rs = 0.625, p = 0.0001), solid appearance (rs = 0.376, p = 0.0001) and hypoechoic appearance (rs = 0.141, p = 0.0001), whereas the detection of malignant thyroid nodules by US was significantly and negatively correlated with well-defined (rs = -0.625, p = 0.0001), cystic (rs = -0.419, p = 0.0001) and isoechoic (rs = -0.146, p = 0.0001) nodules (Table 3). Logistic regression analysis
The predictive value of ultrasonographic features The sensitivity of microcalcification, blurred nodular margins, and solid and hypoechoic appearance in preoperative diagnosis of thyroid carcinoma was 89.3 %, 64.7 %, 81.5%, and 62.6 %, respectively (Figs. 1 and 2).
Logistic regression analysis demonstrated that independent risk factors of malignancy upon US examination of thyroid nodules include microcalcifications (OR: 52.5, p = 0.0001), blurred nodular margins (OR: 16.2, p = 0.0001), solid appearance (OR: 8, p = 0.0001) and hypoechoic appearance
Table 1 Comparative intra-group analysis of ultrasonographic features among benign and malignant thyroid nodules Ultrasonographic features
Malign nodules n = 293
Benign nodules n = 1633
p Value
Size
Largest diameter (mm)
19.8 ± 15
19.2 ± 13
NS
Echogenicity, n; %
Hypoechoic
183; 62.5%
704; 43.1%
0.0001
Isoechoic
89; 30.4%
831; 50.9%
0.0001
Hyperechoic
21; 7.2%
101; 6.2%
NS
Solid
239; 81.6%
449; 30.6%
0.0001
Cystic
0; 0%
995; 58.5%
0.0001
Mixed Microcalcification
54; 18.4% 261; 89%
176; 10.8% 82; 5%
0.0001 0.0001
Coarse
1; 3%
93; 4.8%
0.002
Peripheric
6; 2%
57; 5%
NS
Well-defined
104; 35.5%
1,557; 95.3%
0.0001
Blurred margin
189; 64.5%
76; 4.7%
0.0001
Echo structure, n; %
Calcifications, n; %
Nodule margin, n; %
123
World J Surg (2008) 32:1948–1954
1951
Fig. 1 (A) A hypoechoic solid nodule with well-defined margins. (B) A hypoechoic, ovoid, solid nodule on the left lobe with well-defined margins, slightly extending to the isthmus. (C) An ovoid hyperechoic solid nodule. Note the difference in echogenicity compared to the thyroid gland and the strap muscles. (D) An isoechoic, ovoid solid nodule (arrows) with a thin rim and a few small cystic components
Fig. 2 (A) A large cystic nodule with fine internal septations. (B) A well-defined cystic nodule without internal echogenicities (anechoic). Note the slightly thickened contours. (C) An anechoic cystic nodule with a small, echogenic solid component (arrow). (D) A hypoechoic, ovoid, solid nodule with small anechoic cystic degenerations
123
1952
World J Surg (2008) 32:1948–1954
Fig. 3 (A) A spherical, isoechoic, well-defined nodule with periferal linear calcification. This type of calcification is called ‘‘egg-shell type’’ because of its resemblance to an egg. (B) A large, ovoid nodule with coarse calcifications (arrow) and posterior acoustic shadowing. (C) A large, ovoid, markedly hypoechoic nodule with small echogenic foci due to microcalcifications. Note the lack of a comet-tail and posterior acoustic shadowing. (D) An ovoid cystic nodule with punctate echogenic foci. Note the echogenic tails (arrows). ‘‘The comet-tail’’ sign is suggestive of colloid crystals and should not be mistaken for microcalcifications
Fig. 4 (A) A hypoechoic solid nodule with well-defined margins. (B) A hypoechoic solid nodule with irregular and blurred margins
Table 2 Diagnostic value of ultrasound findings for the detection of malignant thyroid nodules Ultrasonographic features
Sensitivity (%)
Specificity (%)
PPV (%)
NPV (%)
Accuracy (%)
CP (%)
Microcalcification
89.3
94.9
76
98
94.1
76
Blurred margin
64.7*
95.3
71.3
93.7
90.7
71
Solid appearance
81.5
69.4*
32.2*
95.5
71*
32*
Hypoechoic appearance
62.6*
56.9*
20.6*
85.5*
57.8*
20*
* p \ 0.05; comparison with microcalcification PPV positive predictive value; NPV negative predictive value; CP conditional probability
(OR: 1.7, p = 0.02). All the above ultrasonographic features had a 293-fold (OR: 293.9, p = 0.0001) increased rate for detection of malignant thyroid nodules by US
123
(Table 4). Age and nodules greater than 10 mm were not significantly associated with a histological diagnosis of malignancy.
World J Surg (2008) 32:1948–1954
1953
Table 3 Significant correlations between ultrasonographic features Correlations
r
p Value
Malignant nodule/microcalcification
0.791
Malignant nodule/blurred nodule margin
0.625
0.0001
Malignant nodule/solid appearance
0.376
0.0001
Malignant nodule/hypoechoic appearance
0.0001
0.141
0.0001
-0.625
0.0001
Malignant nodule/cystic appearance
-0.419
0.0001
Malignant nodule/hypoechoic appearance
-0.146
0.0001
Malignant nodule/well defined
Table 4 Predictors of histological malignancy according to logistic regression analysis Predictors
OR (95% CI)
p Value
Microcalcification
52.6 (32.5–84.9)
Blurred nodule margin
16.2 (9.2–28.6)
0.0001
8 (4.6–13.9)
0.0001
Solid appearance Hypoechoic appearance All ultrasonographic features
1.7 (1–2.8) 293.9 (–)
0.0001
0.02 0.0001
Discussion We investigated the diagnostic value of ultrasonographic features in detecting malignant nodules in multinodular goiters and determined that microcalcifications, blurred nodular margins, and solid and hypoechoic appearance were significantly associated with thyroid carcinoma independent of tumor size and patient age. Although there is no single criterion that distinguishes benign from malignant thyroid nodules with 100% reliability, microcalcification and blurred nodular margins are useful ultrasonographic features for the detection or the exclusion of thyroid carcinoma with high diagnostic accuracy in patients with multinodular goiter. The high diagnostic accuracy of these features may lead to detection of clinically insignificant nodules and avoid unnecessary work-up and anxiety for the patient. Incidental thyroid nodules are an increasingly common finding because of the expanding use of high-resolution US [3, 5]. Although most of these lesions are benign, clinicians must distinguish these nodules from benign nodules, which can often be managed conservatively. Another problem for clinicians and radiologists is the selection of nodules for FNAB [5, 14, 15]. We are frequently confronted with this problem when it is not possible to biopsy all the nodules. In several studies, the malignancy rate within thyroid incidentalomas was between 13% and 28.8%. In these studies, all nodules were incidentally found, and incidental thyroid carcinoma was diagnosed following FNAB [16–20]. Although many papillary microcarcinomas may remain occult, some
microcarcinomas may result in a negative outcome, including distant metastasis and patient death [19, 20]. Ultrasonography has emerged as a widely used tool in the evaluation of thyroid nodules [1, 3]. Although many sonographic criteria have been developed to allow early and more accurate diagnosis, there are some limitations in sensitivity and specificity [6, 7, 21–23]. Ultrasonography is a useful tool in determining the objective size of a thyroid nodule. However, nodule size does not correlate with malignancy [24]. In our study, 55 of the 293 malignant nodules were less than 10 mm, and the nodule size did not correlate with the detection of malignant thyroid nodules by US. There is a common but mistaken practice of selecting the largest nodule in a multinodular thyroid for FNA. The Society of Radiologists in Ultrasound recently recommended that the selection of a nodule for FNA in a multinodular thyroid be based primarily on US characteristics rather than nodule size [6, 7]. Thyroid calcifications can be classified as microcalcification, coarse calcification, or peripheral calcification [6, 7]. Microcalcifications are defined as punctate hyperechoic foci lacking acoustic shadowing. Pathologically, punctate microcalcifications correlate with the calcification of psammoma bodies. There is almost unanimous agreement that the presence of microcalcifications within a nodule is associated with thyroid carcinoma [10, 11, 20–23]. In our study, the presence of microcalcifications had a higher diagnostic accuracy for malignancy (94.1%) than a solid (71%) or hypoechoic appearance (57.8%) of the nodule. However, coarse calcifications are common and can be found in both benign and malignant nodules. Large, coarse areas of calcification are occasionally seen within a benign nodule as part of a degenerative process [6, 7]. These appear as irregular, hyperechoic foci with an acoustic shadow. Benign nodules may have a peripheral rim of calcification, or ‘‘eggshell calcification.’’ Although peripheral calcification suggests a benign process, it is rarely seen in papillary thyroid carcinomas [6, 7, 10]. The echo texture of a thyroid nodule refers to its echogenicity relative to the background thyroid parenchyma [6–8]. Several studies have mentioned hypoechogenicity as a finding suggestive of malignancy. However, most nonpalpable thyroid nodules are hypoechoic and benign. Nodules can be classified semiquantitatively according to the estimated percentage of solid or cystic composition or in descriptive terms based on the predominant composition [20–23]. Solid or predominantly solid nodules have a higher risk of malignancy than mixed or predominantly cystic nodules. These features have a fairly low positive predictive value, as only 15-27% of solid hypoechoic nodules are malignant [11, 12, 14, 15]. We have also demonstrated that the PPV of solid and hypoechoic nodules was 32.2% and 20.6%, respectively.
123
1954
The contour of a nodule can be described as smooth and round or irregular and angulated. Irregular and angulated contours are thought to be a hallmark of nodules, as the neoplastic thyroid follicular epithelia undergo disorganized growth resulting in a lobulated appearance. The reported sensitivity of ill-defined margins ranges widely (53–89%) [6, 7, 20, 21]. The sensitivities, specificities, and negative and positive predictive values for these criteria are extremely variable from study to study [11–16, 21–24]. The diagnostic value of ultrasonographic features to detect the malignant nodules was evaluated only by retrospective pathologic records of a pool of FNA biopsies of thyroid nodules and did not incorporate the final pathology. The number of the nodules is limited in most studies. The overall predictive value of sonographic findings for thyroid carcinoma was higher in our study compared to previous studies. This finding is probably due to the correlation of sonographic features with the final histopathological results, rather than FNAB results. In other studies, the nodules with benign cytological findings that were not operated were accepted as TN results, whereas these nodules might be actually be FN. This might explain the low diagnostic accuracy of sonographic features for thyroid carcinoma in those studies. The primary aim of our study was to investigate the predictive value of ultrasonographic features in the preoperative diagnosis of malignant thyroid nodules. According to logistic regression analysis, the presence of microcalcifications, blurred nodule margins, and solid and hypoechoic appearance were significant predictive factors for preoperative diagnosis of malignant thyroid nodules. Although there was not a single feature of malignancy in the sonographic examination of a nodule, the presence of microcalcifications and blurred nodule margins had a higher diagnostic accuracy for malignancy compared to other features. These features might be useful for the differentiation of malignant nodules from benign nodules and for the selection of nodules for FNAB, thereby avoiding unnecessary thyroidectomy in patients with MNG.
References 1. Lawrence W Jr, Kaplan BJ (2002) Diagnosis and management of patients with thyroid nodules. J Surg Oncol 80:157–170 2. Bonnema SJ, Bennedbaek FN, Ladenson PW et al (2002) Management of the nontoxic multinodular goiter: a North American survey. J Clin Endocrinol Metab 87:112–117 3. Datta RV, Petrelli NJ, Ramzy J (2006) Evaluation and management of incidentally discovered thyroid nodules. Surg Oncol 15:33–42 4. Jones MK (2001) Management of nodular thyroid disease. The challenge remains identifying which palpable nodules are malignant. BMJ 323:293–294
123
World J Surg (2008) 32:1948–1954 5. Ogilvie JB, Piatigorsky EJ, Clark OH (2006) Current status of fine needle aspiration for thyroid nodules. Adv Surg 40:223–238 6. Frates MC, Benson CB, Charboneau JW et al (2005) Society of Radiologists in Ultrasound. Management of thyroid nodules detected at US: Society of Radiologists in Ultrasound consensus conference statement. Radiology 237:794–800 7. Frates MC, Benson CB, Charboneau JW et al (2006) Management of thyroid nodules detected at US: Society of Radiologists in Ultrasound consensus conference statement. Ultrasound Q 22:231–238 8. Barraclough BM, Barraclough BH (2000) Ultrasound of the thyroid and parathyroid glands. World J Surg 24:158–165 9. Papini E, Guglielmi R, Bianchini A et al (2002) Risk of malignancy in nonpalpable thyroid nodules: predictive value of ultrasound and color Doppler features. J Clin Endocrinol Metab 87:1941–1946 10. Khoo ML, Asa SL, Witterick IJ et al (2002) Thyroid calcification and its association with thyroid carcinoma. Head Neck 24:651– 655 11. Chan BK, Desser TS, McDougall IR et al (2003) Common and uncommon sonographic features of papillary thyroid carcinoma. J Ultrasound Med 22:1083–1090 12. Wienke JR, Chong WK, Fielding JR et al (2003) Sonographic features of benign thyroid nodules: interobserver reliability and overlap with malignancy. J Ultrasound Med 22:1027–1031 13. Alexander EK, Marqusee E, Orcutt J et al (2004) Thyroid nodule shape and prediction of malignancy. Thyroid 14:953–958 14. Jun P, Chow LC, Jeffrey RB (2005) The sonographic features of papillary thyroid carcinomas: pictorial essay. Ultrasound Q 21:39–45 15. Iannuccilli JD, Cronan JJ, Monchik JM (2004) Risk for malignancy of thyroid nodules as assessed by sonographic criteria: the need for biopsy. J Ultrasound Med 23:1455–1464 16. Hoang JK, Lee WL, Lee M et al (2007) US features of thyroid malignancy: pearls and pitfalls. RadioGraphics 27:847–865 17. Ravetto C, Colombo L, Dottorini ME (2000) Usefulness of fineneedle aspiration in the diagnosis of thyroid carcinoma: a retrospective study in 37,895 patients. Cancer 90:357–363 18. Nam-Goong IS, Kim HY, Gong G et al (2004) Ultrasonographyguided fine-needle aspiration of thyroid incidentaloma: correlation with pathological findings. Clin Endocrinol 60:21–28 19. Yokozawa T, Miyauchi A, Kuma K et al (1995) Accurate and simple method of diagnosing thyroid nodules the modified technique of ultrasound-guided fine needle aspiration biopsy. Thyroid 5:141–145 20. Noguchi S, Yamashita H, Murakami N et al (1996) Small carcinomas of the thyroid. A long-term follow-up of 867 patients. Arch Surg 131:187–191 21. Cappelli C, Castellano M, Pirola I et al (2007) The predictive value of ultrasound findings in the management of thyroid nodules. Q J Med 100:29–35 22. Titton RL, Gervais DA, Boland GW et al (2003) Sonography and sonographically guided fine-needle aspiration biopsy of the thyroid gland: indications and techniques, pearls and pitfalls. Am J Roentgenol 181:267–271 23. Koike E, Noguchi S, Yamashita H et al (2001) Ultrasonographic characteristics of thyroid nodules: prediction of malignancy. Arch Surg 136:334–337 24. Kim EK, Park CS, Chung WY et al (2002) New sonographic criteria for recommending fine-needle aspiration biopsy of nonpalpable solid nodules of the thyroid. Am J Roentgenol 178:687– 691
