3407jum1163-1318online_Layout 1 6/22/15 9:05 AM Page 1245
ORIGINAL RESEARCH
A Predictive Model for Selecting Malignant Thyroid Nodules in Patients With Nondiagnostic or Indeterminate Fine-Needle Aspiration Cytologic Findings Kuailu Lin, MD, Yingying Xiang, MD, Li Qiao, MD, Ruoli Liu, MD, Siyang Dong, MD, Xiaohua Zhang, MD Article includes CME test
Objectives—The management of nodules with nondiagnostic or indeterminate fineneedle aspiration cytologic findings remains challenging. This study evaluated the clinical differences, conventional sonographic findings, elastographic findings, and cytologic findings for predicting thyroid malignancy. Methods—A total of 167 patients with a nondiagnostic or indeterminate thyroid fineneedle aspiration cytologic findings were enrolled in this study. The clinicopathologic and sonographic data from the patients were analyzed retrospectively to determine the independent predictive factors for thyroid malignancy. Then a scoring system was designed on the basis of statistically significant predictors. Results—Fine-needle aspiration cytologic findings, Thyroid Imaging Reporting and Data System categorization, and elastographic findings were independent predictive factors for thyroid cancer on multivariate analysis. The index points were statistically significant, with a score higher than 3 favoring malignant nodules with sensitivity of 79.37%, specificity of 85.85%, a positive predictive value of 76.9%, and a negative predictive value of 87.3%.
Received July 18, 2014, from the Departments of Oncology (K.L., Y.X., S.D., X.Z.) and Radiology (L.Q., R.L.), First Affiliated Hospital of Wenzhou Medical University, Wenzhou, China. Revision requested August 7, 2014. Revised manuscript accepted for publication October 16, 2014. This research was supported by National High Technology Research and Development Program 863 (grant 2012 AA02A210). Drs Lin and Xiang contributed equally to this work. Address correspondence to Xiaohua Zhang, MD, Department of Oncology, First Affiliated Hospital of Wenzhou Medical University, 325000 Wenzhou, Zhejiang, China. E-mail: [email protected] Abbreviations
OR, odds ratio; TI-RADS, Thyroid Imaging Reporting and Data System doi:10.7863/ultra.34.7.1245
Conclusions—For patients with nondiagnostic or indeterminate fine-needle aspiration cytologic findings, our scoring system for prediction of thyroid malignancy can be another choice. We suggest surgery for nodules with index points higher than 3. For nodules with index points of 3 or lower, observation and regular follow-up are recommended. Key Words—cytopathology; fine-needle aspiration; head and neck ultrasound; sonography; thyroid cancer; thyroid nodules
D
ue to the wide use of routine high-resolution thyroid sonographic surveys during physical examinations, the incidence of thyroid nodules has been increasing sharply during recent years in eastern China.1 Up to now, ultrasound-guided fine-needle aspiration has been the best single method for differentiating malignant from benign thyroid nodules because of its high sensitivity and specificity.2 The highlighted limitation is that about 30% of fine-needle aspiration cytologic findings remain nondiagnostic or indeterminate.3–6 In patients with nondiagnostic or indeterminate fine-needle aspiration cytologic findings, efforts to improve management have focused on identifying those at highest risk of malignancy.
©2015 by the American Institute of Ultrasound in Medicine | J Ultrasound Med 2015; 34:1245–1251 | 0278-4297 | www.aium.org
3407jum1163-1318online_Layout 1 6/22/15 9:05 AM Page 1246
Lin et al—Predictive Model for Selecting Malignant Thyroid Nodules
The Thyroid Imaging Reporting and Data System (TI-RADS) was developed on the basis of the number of suspicious sonographic features, which has become an accepted way of guiding clinical management of thyroid nodules.7 In this study, we used only the TI-RADS categorization described by Horvath et al7 (Table 1), which shows an excellent negative predictive value for diagnosis of malignant thyroid nodules. Ultrasound elastography has recently been proposed as a dynamic sonographic technique for estimating tissue stiffness by measuring the degree of elasticity under the application of an external force.8 Previous studies have shown that the TI-RADS and elastography are effective tools for diagnosis of thyroid cancer.3,9,10 Herein, we report a series of nondiagnostic or indeterminate fine-needle aspiration cytologic findings, the combination of known risk factors with TI-RADS categorizations and elastographic scores, and the development of a scoring system based on our findings to predict malignancy. The aim of this study was to estimate the probability of malignancy in thyroid nodules with nondiagnostic or indeterminate cytologic findings and assist surgeons in decision making regarding further treatment.
Materials and Methods Patients In this retrospective study, 849 patients underwent ultrasound-guided fine-needle aspiration of thyroid nodules at the First Affiliated Hospital of Wenzhou Medical University
from January 2012 through March 2014. For selection of cases, the selection criteria were as follows: (1) each patient had only 1 nodule with indeterminate or nondiagnostic cytologic findings at 2 different fine-needle aspiration cytologic evaluations; and (2) each patient had undergone thyroid surgery. A total of 167 patients were included in the final analysis. The study was approved by the local hospital Ethics Committee, and informed consent was obtained from each patient. Thyroid Sonography and Elastography All patients were examined by thyroid sonography with Acuson Sequoia and 128XP ultrasound scanners (Siemens Medical Solutions, Mountain View, CA) equipped with 8–13-MHz linear probes. Real-time sonography and subsequent fine-needle aspiration were performed by a single skilled radiologist. Images were saved in picture archiving and communication systems, and the features of the thyroid nodules were recorded. Two experienced ultrasound radiologists, who were blinded to clinical and pathologic data, described the sonographic features and scored the images according to the distribution of the probability of malignancy (Table 1)7 as follows: TI-RADS category 2, benign conditions (0% malignancy); TI-RADS category 3, probably benign nodules (80% malignancy). Elastography was performed with a Hi Vision Preirus system (Hitachi Medical Systems, Tokyo, Japan)
Table 1. Sonographic Characteristics of Thyroid Nodules With Their Malignancy Risk and TI-RADS Categorization Sonographic Characteristics
Malignancy Risk, %
TI-RADS Category
Colloid type 1 nodule: anechoic nodule with hyperechoic spots, nonvascularized nodule Colloid type 2 nodule: nonencapsulated, mixed, nonexpansile nodule with hyperechoic spots, vascularized nodule, spongiform nodule Colloid type 3 nodule: nonencapsulated, mixed nodule with solid portion, isoechoic, expansile, vascularized nodule with hyperechoic spots
0
2, benign
Hashimoto pseudonodule: hyperechoic, isoechoic, or hypoechoic, partially encapsulated nodule with peripheral vascularization
80
5, consistent with malignancy
1246
J Ultrasound Med 2015; 34:1245–1251
3407jum1163-1318online_Layout 1 6/22/15 9:05 AM Page 1247
Lin et al—Predictive Model for Selecting Malignant Thyroid Nodules
by another board-certified radiologist. The elastogram was displayed with an elasticity color scale, ranging from red (components with greatest elastic strain) to blue (components with no strain). According to the elasticity classification modified by Rago et al,3 a score of 1 described nodules with high elasticity (soft); a score of 2 indicated nodules with indeterminate elasticity; and a score of 3 predicted nodules with low elasticity (hard). Both real-time sonography and elastography were performed before the first fine-needle aspiration. Fine-Needle Aspiration Cytologic and Histologic Analyses Ultrasound-guided fine-needle aspiration was performed either on the thyroid nodule with a probability of malignancy that showed suspicious sonographic features (hypoechogenicity, microcalcifications, increased central vascularity, infiltrative margins, and a taller-than-wide shape in the transverse plane) or on the largest nodule if no suspicious nodules were present. Fine-needle aspiration was performed under the sonographic guidance using a 23-gauge needle attached to a 10-mL syringe with an aspirator. In most cases, 2 to 3 fine-needle aspiration passes were performed. The aspirated sample was used to make direct cytologic smears. Specimens were smeared on a slide, fixed in 95% ethanol, and stained by the hematoxylineosin method immediately. Based on features recorded in the pathology database, nondiagnostic or indeterminate cytologic findings were reported by the reading pathologist as follows: (1) Hürthle cell neoplasm; (2) follicular neoplasm; or (3) suspicious for papillary thyroid carcinoma. Patients with nondiagnostic or indeterminate cytologic findings were asked to undergo a second fine-needle aspiration 3 months after the first aspiration. The second aspiration was performed by the same radiologist. Twentyone cases with 2 different cytologic findings were excluded from the study. After thyroid surgery, tissue was fixed in formalin, embedded in paraffin, and stained with hematoxylineosin. The histologic diagnoses were made by 2 senior pathologists in a blinded manner according to the World Health Organization guidelines.11 Statistical Analysis All data were analyzed with SPSS version 19.0 software (IBM Corporation, Armonk, NY). P < .05 was considered statistically significant. For the univariate analysis, we used a χ2 test to analyze the relationship of each variable with the final histopathologic finding (malignant or benign). For the multivariate analysis, a logistic regression model was used to assess independent associations of malignancy
J Ultrasound Med 2015; 34:1245–1251
with all factors that reached statistical significance in the univariate analysis. Finally, a scoring system was set up by giving different points to the independent predictors according to the multiple logistic regression analysis. Moreover, by using receiver-operating characteristic curves, the best point with high sensitivity and a low falsenegative rate (1 – specificity) was identified.
Results Among the 167 patients included in the study, 53 were male, and 114 were female. The mean age of patients at diagnosis was 45.14 (SD, 10.1) years (range, 16–73 years). The mean diameter of the nodules was 1.64 (SD, 1.1) cm (range, 0.4–7.6 cm). Final histologic diagnoses revealed 104 benign nodules (62.3%) and 63 malignant nodules (37.7%).
Table 2. Clinicopathologic and Sonographic Findings According to Pathologic Findings
Characteristic Age, y
ORIGINAL RESEARCH
A Predictive Model for Selecting Malignant Thyroid Nodules in Patients With Nondiagnostic or Indeterminate Fine-Needle Aspiration Cytologic Findings Kuailu Lin, MD, Yingying Xiang, MD, Li Qiao, MD, Ruoli Liu, MD, Siyang Dong, MD, Xiaohua Zhang, MD Article includes CME test
Objectives—The management of nodules with nondiagnostic or indeterminate fineneedle aspiration cytologic findings remains challenging. This study evaluated the clinical differences, conventional sonographic findings, elastographic findings, and cytologic findings for predicting thyroid malignancy. Methods—A total of 167 patients with a nondiagnostic or indeterminate thyroid fineneedle aspiration cytologic findings were enrolled in this study. The clinicopathologic and sonographic data from the patients were analyzed retrospectively to determine the independent predictive factors for thyroid malignancy. Then a scoring system was designed on the basis of statistically significant predictors. Results—Fine-needle aspiration cytologic findings, Thyroid Imaging Reporting and Data System categorization, and elastographic findings were independent predictive factors for thyroid cancer on multivariate analysis. The index points were statistically significant, with a score higher than 3 favoring malignant nodules with sensitivity of 79.37%, specificity of 85.85%, a positive predictive value of 76.9%, and a negative predictive value of 87.3%.
Received July 18, 2014, from the Departments of Oncology (K.L., Y.X., S.D., X.Z.) and Radiology (L.Q., R.L.), First Affiliated Hospital of Wenzhou Medical University, Wenzhou, China. Revision requested August 7, 2014. Revised manuscript accepted for publication October 16, 2014. This research was supported by National High Technology Research and Development Program 863 (grant 2012 AA02A210). Drs Lin and Xiang contributed equally to this work. Address correspondence to Xiaohua Zhang, MD, Department of Oncology, First Affiliated Hospital of Wenzhou Medical University, 325000 Wenzhou, Zhejiang, China. E-mail: [email protected] Abbreviations
OR, odds ratio; TI-RADS, Thyroid Imaging Reporting and Data System doi:10.7863/ultra.34.7.1245
Conclusions—For patients with nondiagnostic or indeterminate fine-needle aspiration cytologic findings, our scoring system for prediction of thyroid malignancy can be another choice. We suggest surgery for nodules with index points higher than 3. For nodules with index points of 3 or lower, observation and regular follow-up are recommended. Key Words—cytopathology; fine-needle aspiration; head and neck ultrasound; sonography; thyroid cancer; thyroid nodules
D
ue to the wide use of routine high-resolution thyroid sonographic surveys during physical examinations, the incidence of thyroid nodules has been increasing sharply during recent years in eastern China.1 Up to now, ultrasound-guided fine-needle aspiration has been the best single method for differentiating malignant from benign thyroid nodules because of its high sensitivity and specificity.2 The highlighted limitation is that about 30% of fine-needle aspiration cytologic findings remain nondiagnostic or indeterminate.3–6 In patients with nondiagnostic or indeterminate fine-needle aspiration cytologic findings, efforts to improve management have focused on identifying those at highest risk of malignancy.
©2015 by the American Institute of Ultrasound in Medicine | J Ultrasound Med 2015; 34:1245–1251 | 0278-4297 | www.aium.org
3407jum1163-1318online_Layout 1 6/22/15 9:05 AM Page 1246
Lin et al—Predictive Model for Selecting Malignant Thyroid Nodules
The Thyroid Imaging Reporting and Data System (TI-RADS) was developed on the basis of the number of suspicious sonographic features, which has become an accepted way of guiding clinical management of thyroid nodules.7 In this study, we used only the TI-RADS categorization described by Horvath et al7 (Table 1), which shows an excellent negative predictive value for diagnosis of malignant thyroid nodules. Ultrasound elastography has recently been proposed as a dynamic sonographic technique for estimating tissue stiffness by measuring the degree of elasticity under the application of an external force.8 Previous studies have shown that the TI-RADS and elastography are effective tools for diagnosis of thyroid cancer.3,9,10 Herein, we report a series of nondiagnostic or indeterminate fine-needle aspiration cytologic findings, the combination of known risk factors with TI-RADS categorizations and elastographic scores, and the development of a scoring system based on our findings to predict malignancy. The aim of this study was to estimate the probability of malignancy in thyroid nodules with nondiagnostic or indeterminate cytologic findings and assist surgeons in decision making regarding further treatment.
Materials and Methods Patients In this retrospective study, 849 patients underwent ultrasound-guided fine-needle aspiration of thyroid nodules at the First Affiliated Hospital of Wenzhou Medical University
from January 2012 through March 2014. For selection of cases, the selection criteria were as follows: (1) each patient had only 1 nodule with indeterminate or nondiagnostic cytologic findings at 2 different fine-needle aspiration cytologic evaluations; and (2) each patient had undergone thyroid surgery. A total of 167 patients were included in the final analysis. The study was approved by the local hospital Ethics Committee, and informed consent was obtained from each patient. Thyroid Sonography and Elastography All patients were examined by thyroid sonography with Acuson Sequoia and 128XP ultrasound scanners (Siemens Medical Solutions, Mountain View, CA) equipped with 8–13-MHz linear probes. Real-time sonography and subsequent fine-needle aspiration were performed by a single skilled radiologist. Images were saved in picture archiving and communication systems, and the features of the thyroid nodules were recorded. Two experienced ultrasound radiologists, who were blinded to clinical and pathologic data, described the sonographic features and scored the images according to the distribution of the probability of malignancy (Table 1)7 as follows: TI-RADS category 2, benign conditions (0% malignancy); TI-RADS category 3, probably benign nodules (80% malignancy). Elastography was performed with a Hi Vision Preirus system (Hitachi Medical Systems, Tokyo, Japan)
Table 1. Sonographic Characteristics of Thyroid Nodules With Their Malignancy Risk and TI-RADS Categorization Sonographic Characteristics
Malignancy Risk, %
TI-RADS Category
Colloid type 1 nodule: anechoic nodule with hyperechoic spots, nonvascularized nodule Colloid type 2 nodule: nonencapsulated, mixed, nonexpansile nodule with hyperechoic spots, vascularized nodule, spongiform nodule Colloid type 3 nodule: nonencapsulated, mixed nodule with solid portion, isoechoic, expansile, vascularized nodule with hyperechoic spots
0
2, benign
Hashimoto pseudonodule: hyperechoic, isoechoic, or hypoechoic, partially encapsulated nodule with peripheral vascularization
80
5, consistent with malignancy
1246
J Ultrasound Med 2015; 34:1245–1251
3407jum1163-1318online_Layout 1 6/22/15 9:05 AM Page 1247
Lin et al—Predictive Model for Selecting Malignant Thyroid Nodules
by another board-certified radiologist. The elastogram was displayed with an elasticity color scale, ranging from red (components with greatest elastic strain) to blue (components with no strain). According to the elasticity classification modified by Rago et al,3 a score of 1 described nodules with high elasticity (soft); a score of 2 indicated nodules with indeterminate elasticity; and a score of 3 predicted nodules with low elasticity (hard). Both real-time sonography and elastography were performed before the first fine-needle aspiration. Fine-Needle Aspiration Cytologic and Histologic Analyses Ultrasound-guided fine-needle aspiration was performed either on the thyroid nodule with a probability of malignancy that showed suspicious sonographic features (hypoechogenicity, microcalcifications, increased central vascularity, infiltrative margins, and a taller-than-wide shape in the transverse plane) or on the largest nodule if no suspicious nodules were present. Fine-needle aspiration was performed under the sonographic guidance using a 23-gauge needle attached to a 10-mL syringe with an aspirator. In most cases, 2 to 3 fine-needle aspiration passes were performed. The aspirated sample was used to make direct cytologic smears. Specimens were smeared on a slide, fixed in 95% ethanol, and stained by the hematoxylineosin method immediately. Based on features recorded in the pathology database, nondiagnostic or indeterminate cytologic findings were reported by the reading pathologist as follows: (1) Hürthle cell neoplasm; (2) follicular neoplasm; or (3) suspicious for papillary thyroid carcinoma. Patients with nondiagnostic or indeterminate cytologic findings were asked to undergo a second fine-needle aspiration 3 months after the first aspiration. The second aspiration was performed by the same radiologist. Twentyone cases with 2 different cytologic findings were excluded from the study. After thyroid surgery, tissue was fixed in formalin, embedded in paraffin, and stained with hematoxylineosin. The histologic diagnoses were made by 2 senior pathologists in a blinded manner according to the World Health Organization guidelines.11 Statistical Analysis All data were analyzed with SPSS version 19.0 software (IBM Corporation, Armonk, NY). P < .05 was considered statistically significant. For the univariate analysis, we used a χ2 test to analyze the relationship of each variable with the final histopathologic finding (malignant or benign). For the multivariate analysis, a logistic regression model was used to assess independent associations of malignancy
J Ultrasound Med 2015; 34:1245–1251
with all factors that reached statistical significance in the univariate analysis. Finally, a scoring system was set up by giving different points to the independent predictors according to the multiple logistic regression analysis. Moreover, by using receiver-operating characteristic curves, the best point with high sensitivity and a low falsenegative rate (1 – specificity) was identified.
Results Among the 167 patients included in the study, 53 were male, and 114 were female. The mean age of patients at diagnosis was 45.14 (SD, 10.1) years (range, 16–73 years). The mean diameter of the nodules was 1.64 (SD, 1.1) cm (range, 0.4–7.6 cm). Final histologic diagnoses revealed 104 benign nodules (62.3%) and 63 malignant nodules (37.7%).
Table 2. Clinicopathologic and Sonographic Findings According to Pathologic Findings
Characteristic Age, y
