European Journal of Radiology 84 (2015) 407–412

Contents lists available at ScienceDirect

European Journal of Radiology journal homepage: www.elsevier.com/locate/ejrad

Shear wave elastography of thyroid nodules for the prediction of malignancy in a large scale study Ah Young Park a , Eun Ju Son a , Kyunghwa Han b , Ji Hyun Youk a , Jeong-Ah Kim a,∗ , Cheong Soo Park c a

Department of Radiology, Yonsei University College of Medicine, Gangnam Severance Hospital, Seoul, Republic of Korea Biostatistics Collaboration Unit, Gangnam Medical Research Center, Yonsei University College of Medicine, Gangnam Severance Hospital, Seoul, Republic of Korea c Department of Surgery, Yonsei University College of Medicine, Gangnam Severance Hospital, Seoul, Republic of Korea b

a r t i c l e

i n f o

Article history: Received 31 July 2014 Received in revised form 14 November 2014 Accepted 18 November 2014 Keywords: Shear wave elastography Thyroid nodule Malignancy

a b s t r a c t Objectives: The purpose of this study is to validate the usefulness of shear wave elastography (SWE) in predicting thyroid malignancy with a large-scale quantitative SWE data. Methods: This restrospective study included 476 thyroid nodules in 453 patients who underwent grayscale US and SWE before US-guided fine-needle aspiration biopsy (US-FNA) or surgical excision were included. Gray-scale findings and SWE elasticity indices (EIs) were retrospectively reviewed and compared between benign and malignant thyroid nodules. The optimal cut-off values of EIs for predicting malignancy were determined. The diagnostic performances of gray-scale US and SWE for predicting malignancy were analyzed. The diagnostic performance was compared between the gray-scale US findings only and the combined use of gray-scale US findings with SWEs. Results: All EIs of malignant thyroid nodules were significantly higher than those of benign nodules (p ≤ .001). The optimal cut-off value of each EI for predicting malignancy was 85.2 kPa of Emean , 94.0 kPa of Emax , 54.0 kPa of Emin . Emean (OR 3.071, p = .005) and Emax (OR 3.015, p = .003) were the independent predictors of thyroid malignancy. Combined use of gray-scale US findings and each EI showed elevated sensitivity (95.0–95.5% vs 92.9%, p ≤ .005) and AUC (0.820–0.834 vs 0.769, p ≤ .005) for predicting malignancy, compared with the use of only gray-scale US findings. Conclusions: Quantitative parameters of SWE were the independent predictors of thyroid malignancy and SWE evaluation combined with gray-scale US was adjunctive to the diagnostic performance of gray-scale US for predicting thyroid malignancy. © 2014 Elsevier Ireland Ltd. All rights reserved.

1. Introduction High-resolution gray-scale ultrasound (US) has played an important role in differentiating benign from malignant thyroid nodules [1]. Reported sonographic characteristics that are suggestive of thyroid malignancy include the following findings: microcalcification, irregular or microlobulated margin, marked hypoechogenicity, and taller-than-wide shape [2]. However, each characteristic has relatively low sensitivity (26.5–59.1%) for predicting thyroid malignancy and there is still considerable overlap of US findings in benign and malignant nodules [1,3,4]. ∗ Corresponding author at: Department of Radiology, Gangnam Severance Hospital, Yonsei University College of Medicine, 211 Eonjuro, Gangnam-gu, Seoul 135-720, South Korea. Tel.: +82 2 2019 3510; fax: +82 2 3462 5472. E-mail address: [email protected] (J.-A. Kim). http://dx.doi.org/10.1016/j.ejrad.2014.11.019 0720-048X/© 2014 Elsevier Ireland Ltd. All rights reserved.

Recently, US elastography has emerged as a new adjunctive tool for gray-scale US by evaluating the tissue stiffness [5–7]. The mechanism of US elastography is measuring the degree of tissue distortion under the application of an external force, either freehand compression or mechanical vibration created by a focused ultrasound beam [5–7]. The usefulness of US elastography to evaluate breast lesions has already been acknowledged through many recent studies, which showed that malignant breast lesions are harder than the benign lesions [8]. However, the usefulness of US elastography to evaluate thyroid lesions is still controversial and most of the recent studies were performed with the static compression elastography which has limitations due to lack of quantitative assessment of tissue stiffness and low reproducibility from operator dependency [9–20]. Shear wave elastography (SWE) is a newly developed quantitative elastography which uses mechanical vibrations through

408

A.Y. Park et al. / European Journal of Radiology 84 (2015) 407–412

acoustic radiation force created by a focused US beam. The displacement induced at the focus generates a shear wave which conveys information linked to the local viscoelastic properties of the tissue, thus enabling a quantitative approach to elasticity values [21]. This new technique depends less on the individual operator and is more reproducible and quantitative [8]. There have been several reports about SWE for the quantitative evaluation of thyroid nodules elasticity [22–27]. These reports have documented that quantitative parameters of elastography were significantly different between malignant and benign nodules. However, the optimal cut-off values of elasticity index (EI) for predicting malignancy showed a wide range from 35 to 66 kPa with variable diagnostic performances and the reference standard has not been established [22–26]. Therefore, the purpose of this study was to validate the usefulness of SWE in predicting thyroid malignancy with largescale quantitative SWE data and to establish the reference value of EI for predicting thyroid malignancy.

obtaining gray-scale US, SWE images were obtained for the thyroid nodules on longitudinal plane by using a freehand technique. The built-in region-of-interest (ROI) (Q-box; Super Sonic Imagine) of the system was set to include the lesion and the surrounding normal tissue, which demonstrated the semitransparent color map of tissue stiffness overlaid gray-scale image with a range from dark blue, indicating the lowest stiffness up to red, indicating the highest stiffness (0–180 kPa). Fixed 1 × 1 or 2 × 2-mm ROIs were placed by an investigator over the stiffest part of the lesion with the exception of calcification. The system calculated elasticity indices (EIs) of Emean (mean value for the ROI placed in the stiffest region), Emax (maximum pixel value for the ROI placed in the stiffest region), and Emin (minimum pixel value for the ROI placed in the stiffest region) (Fig. 1). The second ROI of a same size was placed in the normal thyroid parenchyma and anterior strap muscle (Fig. 1). The elastic ratios of mean stiffness for the lesion-to-normal parenchyma (Emean-p ) and the lesion-to-strap muscle (Emean-m ) were calculated.

2. Materials and methods The institutional review board approved this retrospective observational study and required neither patient approval nor informed consent for review of patient images and records. However, informed consent was obtained from all patients for USguided fine-needle aspiration biopsy (US-FNA) and surgery before each procedure. 2.1. Patients Between March 2012 and December 2012, consecutive 493 patients with 521 thyroid nodules underwent gray-scale US and SWE before US-FNA (n = 123) or surgical excision (n = 398) at our institute. Among these lesions, 45 lesions in 40 patients were excluded in this study because non-diagnostic results were yielded after US-FNA (Bethesda category I) (n = 18) or subsequent surgery was not performed for the lesions which were diagnosed as indetermined or suspicious for malignancy with US-FNA (Bethesda category III through VI, n = 27) [28]. A total 476 thyroid nodules in 453 patients were pathologically confirmed as either benignity or malignancy, and enrolled in this study. Patients included 387 women and 89 men (mean age, 45.7 years; range, 15–77 years).

2.3. Histopathologic diagnosis Ninety-seven thyroid nodules were benign and 379 thyroid nodules were malignant. Benignity was diagnosed either by US-FNA or by surgical excision. The cytologic results were reported according to the Bethesda system for reporting thyroid cytopathology [28]. The benign lesions were diagnosed with US-FNA when benign results were obtained at 2 consecutive US-FNAs and there were no interval changes on follow-up US in more than a year, and further surgical excision was not performed (n = 73). Remaining 24 thyroid nodules were diagnosed as benign after surgical excision due to the presence of concurrent thyroid malignancy or the US-FNA results of category III or IV. All malignancies were diagnosed by surgical excision. Total or hemi-thyroidectomy was performed for the lesions diagnosed as category III through VI at US-FNA (n = 363). Other 16 malignant lesions were excised without US-FNA due to the presence of concurrent thyroid malignancy. After fixation, embedding and staining, tissue sections were examined by two pathologists with 20–25 years of experience in thyroid cytopathology. Histological typing was based on the World Health Organization (WHO) guideline [29]. 2.4. Statistical analysis

2.2. US examination Thyroid US examinations were performed with Aixplorer US system (Super Sonic Imagine, Aix-en-Provence, France) equipped with a 15–4-MHz linear-array transducer by two radiologists with 10–15 years of experience in thyroid US. Findings of gray-scale US and SWE were prospectively recorded in the radiologic reports and our database. A careful evaluation of the following gray-scale US findings was performed: echogenicity (hyperechoic, isoechoic, hypoechoic or marked hypoechoic), margin (circumscribed, microlobulated or irregular), calcification (microcalcification, macrocalcification, eggshell calcification, mixed calcification-combined type of microcalcification with any other type of calcification or negative), and shape (wider-than-tall or taller-than-wide) [2]. Color-flow Doppler US was also performed to examine the vascularity of lesion (negative, central, peripheral or mixed vascularity-both central and peripheral vascularity in one lesion). Thyroid lesions showing one or more of the following suspicious gray scale US findings were regarded as suspicious category for malignancy: marked hypoechogenicity, irregular margin, microcalcification or mixed calcification, or taller-than-wide shape [2]. Thyroid lesions without any suspicious findings were regarded as probably benign. After

Gray-scale US findings were compared using Chi-square test and EIs were compared using Student’s t-test between benign and malignant thyroid nodules. Receiver operating characteristic curve analysis with the use of Youden’s index was performed to determine the optimal cut-off values of EIs for predicting malignancy and their diagnostic performance. Logistic regression analysis was performed to evaluate the power for predicting malignancy of each suspicious gray-scale US finding and each optimal cut-off value of EI. Comparison of diagnostic performance for predicting malignancy between the gray-scale US findings only and the combined gray-scale US findings and the EIs was performed by logistic regression analysis with generalized estimating equation method. Diagnostic performance of gray-scale US findings to predict malignancy was calculated on the base of assumption that it is suspicious for malignancy if the thyroid nodule has one or more suspicious finding. Comparison of the area under the curve (AUC) between the gray-scale finding only and the combined use of gray-scale finding and EIs was performed by the method of Delong et al. [30]. Statistical analysis was carried out using SAS (version 9.2, SAS Institute Inc., Cary, NC, USA). For all analyses, twotailed p-values of less than .05 were considered to be statistically significant.

A.Y. Park et al. / European Journal of Radiology 84 (2015) 407–412

409

Fig. 1. A 48-year-old woman with surgically proven papillary thyroid carcinoma. Gray-scale US shows a hypoechoic nodule with irregular margin which was assessed as a suspicious nodule on gray scale US. SWE displays heterogeneous color elasticity signal with peripheral high SWE areas (red). Measured elasticity indices of Emean of 162.9 kPa, Emin of 113.7 kPa and Emax of 184.8 kPa were obtained.

3. Result 3.1. Histopathologic diagnosis Ninety-seven thyroid nodules were benign and 379 thyroid nodules were malignant. Benign thyroid nodules included 2 follicular adenoma, 15 adenomatous hyperplasia, 7 lymphocytic thyroiditis with surgical confirmation and 73 Bethesda category II lesions with US-FNAs. Malignant thyroid nodules included 375 papillary thyroid carcinoma (PTC): 354 conventional subtype, 16 follicular variant, 4 diffuse sclerosing variant, and 1 micropapillary and hobnail variant, 2 follicular carcinoma and 2 medullary carcinoma. 3.2. Gray-scale US findings Suspicious gray-scale US findings and suspicious category were more frequently found in malignant thyroid nodules than in benign nodules with statistical significance (all, p < .0001): marked hypoechogenicity, 236/379 (62.3%) vs 27/97 (27.8%); irregular margin, 235/379 (94.7%) vs 20/97 (20.6%); microcalcification or mixed calcification, 292/379 (77.0%) vs 8/97 (8.3%); taller-than-wide shape, 189/379 (49.9%) vs 13/97 (8.7%); suspicious category, 377/379 (99.5%) vs 42/97 (43.3%). There was no significant difference of vascularity between benign and malignant nodules (p = .605).. 3.3. SWE findings The mean values of all elasticity indices (EI) between benign and malignant thyroid nodules were significantly different with higher values in malignant nodules: Emean , 56.09 ± 26.47 kPa vs 87.84 ± 50.55 kPa, p < .0001; Emax , 65.57 ± 31.13 kPa vs p < .0001; Emin , 44.70 ± 24.11 kPa vs 103.00 ± 61.84 kPa, 66.72 ± 39.76 kPa, p < .0001; Emean-p , 2.86 ± 2.45 vs 4.63 ± 9.59, p = .002; Emean-m , 1.94 ± 1.24 vs 2.85 ± 4.72, p = .001 (Table 1).

The optimal cut-off values of EIs for predicting malignancy were as followed: Emean , 85.2 kPa; Emax , 94.0 kPa; Emin , 54.0 kPa; Emean-p , 2.5; Emean-m , 2.6 (Table 2). Diagnostic performance of each EI with optimal cut-off values for predicting malignancy are summarized in Table 2.

3.4. Predictors of thyroid malignancy On univariate logistic regression analysis, each suspicious gray-scale US finding and each cut-off value of each EI were statistically significant predictors of thyroid malignancy: marked hypoechogenicity, odds ratio [OR] 4.278, p < .0001; irregular margin, OR 6.283, p < .0001; microcalcification or mixed calcification, OR 37.339, p < .0001; taller-than-wide shape, OR 6.428, p < .0001; Emean , OR 6.028, p < .0001; Emax , OR 5.14, p < .0001; Emin , OR 2.934, p < .0001; Emean-p , OR 1.754, p = .016; Emean-m , OR 2.461, p = .002 (Table 3). On multivariate analysis, microcalcification or mixed calcification (OR 23.791–31.080, p < .0001), taller-than-wide shape (OR 3.73–4.554, p < .001), Emean (OR 3.071, p = .005) and Emax (OR

Table 1 Comparison of elasticity indices of shear wave elastography between benign and malignant thyroid nodules.

Emean Emax Emin Emean-p Emean-m

Benigna

Malignanta

p-value

56.09 ± 26.47 65.57 ± 31.13 44.70 ± 24.11 2.86 ± 2.45 1.94 ± 1.24

87.84 ± 50.55 103.00 ± 61.84 66.72 ± 39.76 4.63 ± 9.59 2.85 ± 4.72

Shear wave elastography of thyroid nodules for the prediction of malignancy in a large scale study.

The purpose of this study is to validate the usefulness of shear wave elastography (SWE) in predicting thyroid malignancy with a large-scale quantitat...
603KB Sizes 0 Downloads 6 Views