The Utility of Ultrasound Elastography and MicroPure Imaging in the Differentiation of Benign and Malignant Thyroid Nodules, Nazan çiledag, Kemal Arda, Bilgin Kadri Arıbas, Elif Aktas and Serdal Kenan Köse, AJR: 198, March 2012
Abstract
OBJECTIVE. The aim of this study was to evaluate the utility of ultrasound elastography and MicroPure imaging in the differential diagnosis of benign and malignant thyroid nodules.
SUBJECTS AND METHODS. A total of 74 consecutive patients (65 women and nine men; age range, 21–80 years; mean [± SD] age, 51 ± 12.7 years) with thyroid nodules, who were referred for fine-needle aspiration biopsy by endocrinology or general surgery clinics, were prospectively examined using B-mode ultrasound, ultrasound elastography, and MicroPure imaging. The strain value ratio (strain index) of thyroid nodules was calculated. Patients with malignant or intermediate fine-needle aspiration biopsy results underwent thyroid surgery.
RESULTS. Using MicroPure imaging, 17 of 65 benign thyroid nodules (26.6%) and three of nine malignant thyroid nodules (33.3%) were found to contain microcalcifications. The sensitivity, specificity, negative predictive value, positive predictive value, and the accuracy rate of MicroPure imaging were 42.9%, 80.6%, 93.1%, 18.8%, and 77%, respectively. By using receiver operating characteristic analysis, the best cutoff point (2.31) was computed (area under the curve, 0.87; p < 0.001). The sensitivity, specificity, negative predictive value, positive predictive value and accuracy rate of the strain index values were 85.7%, 82.1%, 98.2%, 33.3%, and 82.4%, respectively, when the best cutoff point of 2.31 was used (p = 0.001). The p value (x = malign) was 0.96 for a strain index value higher than 2.31.
CONCLUSION. This preliminary study indicated that ultrasound elastography and MicroPure imaging can be used for the differentiation of benign and malignant thyroid nodules.
Thyroid nodules are a common finding in the general population, especially in geographic areas of iodine deficiency. Over 95% of thyroid nodules are benign and less than 5% are malignant [1]. Ultrasound is a noninvasive and easily available imaging technique for the evaluation of thyroid nodules. Many studies have reported the utility of ultrasound for the differentiation of benign and malignant thyroid nodules [1–5]. The presence of calcification, hypoechogenicity, irregular margins, absence of a halo, and predominant solid composition in the sonographic image are the key features associated with an increased risk of malignancy. However, the sensitivity, specificity, and negative and positive predictive values for these features are highly variable across patients and across different machines, and no single sonographic feature can diagnose thyroid cancer with high sensitivity and high positive predictive value [1–5].
Fine-needle aspiration biopsy of thyroid lesion is the preoperative screening method of choice worldwide, because it distinguishes benign and malignant lesions with high accuracy [6–8]. Because of its simplicity, low cost, and absence of major complications, it is the initial investigative technique in the management of thyroid diseases [6–8].
Real-time sonographic elastography is a newly developed dynamic imaging technique that displays tissue elasticity by measuring the degree of distortion under the application of an external force. Like palpation, sonographic elastography uses tissue deformation or strain that is caused by external compression and is based on the precompression and compression. Ultrasonographic elastography has been used to examine such organs as the breast [9, 10], thyroid [11], prostate [12], cervix [13], and liver [14]. This technique is a promising imaging technique that can be used for the differentiation of benign and malignant thyroid nodules. However, to our knowledge, only a limited number of studies have described the application of real-time sonographic elastography on benign and malignant thyroid nodules [15].
The MicroPure imaging algorithm (Toshiba) is an adapted filter that is used to enhance bright echoes to visualize and show calcifications, particularly microcalcifications. The purpose of this study was to evaluate the utility of ultrasound elastography and MicroPure imaging in differentiating benign and malignant thyroid nodules.
Subjects and Methods
This prospective study was approved by the Ankara Oncology Research and Education Hospital review board. Written informed consent was obtained from all patients undergoing both real-time ultrasound elastography and MicroPure imaging.
From February 2010 to April 2010, 74 consecutive patients (65 women and nine men; age range, 21–80 years; mean [± SD] age, 51 ± 12.7 years) with incompletely diagnosed thyroid nodules referred for fine-needle aspiration biopsy by endocrinology or general surgery clinics were examined prospectively. Patients with nodules larger than 40 mm, purely cystic or anechoic nodules without solid components, and shell-calcified nodules that could cause color-coding problems were excluded from the study.
All patients were examined by using gray-scale ultrasound, MicroPure imaging, and real-time sonographic elastography with a 10-MHz linear transducer (Aplio, Toshiba) during the same examination by the same operator. In sonographic elastography, the deflections occurring before and after tissue compression were calculated semiquantitatively via the shear modulus (Young modulus) and were displayed graphically in the elastogram.
The gray-scale sonography and elastography in all patients was performed by the same radiologist to prevent differences among operators and to standardize the degree of nodule pressure. All interpretations were performed before biopsy by the same operator, and the radiologist was blinded to the final diagnosis of the patients.
The sonographic examinations were performed in two steps. Gray-scale sonography and MicroPure imaging were performed for all patients in the first step, and real-time sonographic elastography was performed in the second step using the same probe during the same examination. B-mode ultrasound was performed first, then in the second step MicroPure imaging was performed, and the third step was real-time sonographic elastography.
For real-time sonographic elastography, compression was performed repeatedly in a vertical direction with light pressure and was followed by decompression. The strain value ratio (strain index) of thyroid nodule to muscle was calculated. Acquiring measurements at the same depth of the nodule and adjacent muscle was a critical issue for strain ratio calculations. Color coding of elastographic images was classified into five groups according to the Ueno classification [9]. A score of 1 indicated strain for the entire lesion (i.e., the entire lesion was evenly shaded in green) (Fig. 1); a score of 2 indicated strain in most of the lesion with some areas of no strain (i.e., a mosaic pattern of green and blue) (Fig. 2); a score of 3 indicated strain at the periphery of the lesion, with sparing of the center of the lesion (i.e., the peripheral part of the lesion was green, and the central part was blue) (Fig. 3); a score of 4 indicated no strain in the entire lesion (i.e., the entire lesion was blue, but its surrounding area was not included) (Fig. 4); and a score of 5 indicated no strain in the entire lesion or in the surrounding area (i.e., both the entire lesion and its surrounding area were blue) (Fig. 5).
Fig. 1: 56-year-old woman. Elastogram of benign thyroid nodule revealed elastographic color score of 1.
All of the patients underwent fine-needle aspiration biopsy under ultrasound guidance within 5 days of the real-time sonographic elastographic evaluation. A 21-gauge needle was used with an attached 20-mL syringe for fine-needle aspiration biopsies. The procedure was repeated one or two times. The collected material was placed onto glass slides, smeared, and equally fixed in air and 95% ethyl alcohol. Air-dried slides were stained with May-Grünwald-Giemsa stain; alcohol-fixed slides were stained using Papanicolaou method and H and E stain. The cytologic diagnoses of the thyroid nodules were compared with real-time sonographic elastography and the MicroPure imaging features. Nine patients with malignant nodules and eight patients with benign nodules in whom the diagnosis was achieved by cytologic examination underwent surgery, and the pathologic diagnoses were confirmed. Cytologically, 57 benign nodules were monitored by ultrasound for 12 months, and repeated biopsies were performed.
For the statistical analysis, quantitative variables were compared using Student t test for independent samples test, and qualitative variables were compared using the chi-square test. To determine the best of cutoff point for strain index, the receiver operating characteristic analysis was used. The quantitative data are presented as mean (± SD). A p value less than 0.05 indicated statistical significance with a 95% confidence level. All statistical analyses were performed using the SPSS packet program (version 11.5, SPSS).
Results
Seventy-four patients (age range, 21–80 years; mean age, 51 ± 12.7 years), including 65 women (age range, 21–80 years; mean age, 49.9 ± 12.3 years) and nine men (age range, 37–78 years; mean age, 58.2 ± 13.2 years), were enrolled in the study (Table 1). The maximum diameters of the evaluated thyroid nodules were 7–36 mm (mean, 17.2 ± 7.2 mm) (Table 1). Table 1 and Table 2 show the features of the patients and the benign and malignant nodules, respectively.
At ultrasound examination, 74 nodules were identified. Among 74 nodules, 56 nodules (75.7%) were solid, and 18 nodules (24.3%) were heterogeneous with cystic degeneration. Thirty-two nodules (43.2%) were hypoechoic, eight nodules (10.8%) were isoechoic, six nodules (8.1%) were hyperechoic, and 10 nodules (13.5%) were hypoechoic. Regular margins were observed in 60 nodules (81.08%), and irregular margins were seen in 14 nodules (18.9%). Calcifications were found in 17 of 65 benign nodules (26.6%) and in three of nine (33.3%) malignant nodules (Table 2). There was no statistically significant difference (p > 0.05).
Fig. 2: 32-year-old man. Elastogram of benign thyroid nodule showed elastographic color score of 2.
Fig. 3 : 48-year-old man. Elastogram of benign thyroid nodule showed elastographic color score of 3.
Fig. 4 : 56-year-old woman. Elastogram of thyroid nodule revealed elastographic color score of 4. Final diagnosis was papillary carcinoma.
Fig. 5 :
65-year-old woman. Elastogram of thyroid nodule with elastographic color score of 5. Final diagnosis was papillary carcinoma.
Fig. 6: 58-year-old woman. MicroPure (Toshiba) image of thyroid nodule showing microcalcifications. After fine-needle aspiration biopsy, diagnosis in this case was papillary carcinoma.
Fig. 7: 45-year-old woman. Elastogram showed thyroid nodule in right isthmic portion with elasticity index of 5.44, with color score of 4. After fine-needle aspiration biopsy, final diagnosis was papillary carcinoma.
Using MicroPure imaging, 17 of 65 (26.6%) benign thyroid nodules and three of nine (33.3%) malignant thyroid nodules revealed microcalcifications (Fig. 6). The sensitivity, specificity, negative predictive value, positive predictive value, and accuracy rate of MicroPure imaging was 42.9%, 80.6%, 93.1%, 18.8%, and 77%, respectively.
TABLE 1: Demographic Features of Patients Included in This Study
TABLE 2 : Characteristics of Benign and Malignant Nodules
Color coding of elastographic images was classified into five groups according to the Ueno classification [9]; of the 65 benign nodules, 41 nodules (63%) had a score of 1 or 2 (15 nodules had a score of 1, and 26 nodules had a score of 2). Of the nine malignant nodules, eight (88.8%) had a score of 4 or 5 (four nodules had a score of 4, and four nodules had a score of 5). Of the 65 benign nodules, 21 (32.3%) nodules had a score of 3. Only one of the malignant lesions (11.1%) had a color score of 3, and none of the malignant nodules had a color score of 1 or 2. Of the 65 benign nodules, two (3.0%) had a color score of 4 and one (1.5%) had a color score of 5. The sensitivity, specificity, negative predictive value, positive predictive value, and accuracy rate of strain index values were 85.7%, 82.1%, 98.2%, 33.3%, and 82.4%, respectively, when the best cutoff point of 2.31 was used (area under the curve, 0.87; p < 0.001) (Figs. 7 and 8). The p value (x = malign) was 0.96 for strain index values higher than 2.31.
Fig. 8: Results of receiver operating characteristic analysis for strain index. Diagonal segments are produced by ties. With cutoff value of 2.31, area under the curve is 0.87 ± 0.05, asymptotic 95% CI is 0.774– 0.970, and asymptotic significance level is 0.001.
Discussion
Palpation, which is one of the oldest clinical skills, provides information about the stiffness of soft tissues using external compression for physical deformation of the tissue. However, palpation is a subjective examination technique. Elasticity measurements and stiffness evaluations of soft tissues are useful in the differential diagnosis of tumor, inflammation, and normal tissue. It is generally accepted that benign soft-tissue lesions are firmer than normal tissue but softer than cancers [16–19].
Until now, the stiffness of thyroid nodules has not been an objective indicator of malignancy, although it is an important factor in the differential diagnosis of malignant nodules [16]. Although fine-needle aspiration biopsy is the best method for this purpose, it suffers the limitations of being invasive, and sampling errors are inevitable [19]. A recently developed promising imaging technique called real-time sonographic elastography reveals the physical properties of soft tissue by characterizing the difference in elasticity between the region of interest and the surrounding normal soft tissue using manual compression and deformation of the tissue. Essentially, sonographic elastography is based on the combined visualization of tissue elasticity (strain) and the velocity at which the tissue deformation occurs [20]. The degree of deformation of the soft tissue is calculated and combined with a gray-scale ultrasound image as an elastography map for the evaluation of tissue stiffness after ultrasound examination. Therefore, no extra time is needed to perform sonographic elastography.
Some of the major advantages of realtime sonographic elastography are its ease of performance, its noninvasiveness, and its suitability of use during routine ultrasound examinations. In addition, this imaging technique facilitates the dynamic visualization of lesions during compression.
Lyshchik et al. [11] prospectively evaluated the sensitivity and specificity of sonographic elastography for differentiating benign and malignant tumors of the thyroid gland. They reported that thyroid lesions, such as cysts and benign and malignant nodules, exhibit different elastographic characteristics. Cysts appear as dark lesions on elastograms, whereas solid nodular lesions are stiffer than thyroid gland tissue, and malignant lesions are significantly stiffer than benign thyroid nodules. Lyshchik et al. [11] have suggested that a strain index value greater than four is the strongest independent predictor of thyroid gland malignancy (p < 0.001) and exhibits 96% specificity and 82% sensitivity.
Kagoya et al. [19] set a strain ratio or strain index value greater than 1.5 as a predictor of thyroid malignancy. This criterion exhibits 90% sensitivity and 50% specificity. In our study, the sensitivity, specificity, and accuracy rates of the strain index values were 85.7%, 82.1%, and 82.4%, respectively, when the best cutoff point of 2.31 was used.
Dighe et al. [21] studied the differential diagnoses of thyroid nodules with ultrasound elastography using carotid artery pulsation as a compression source combined with limited external compression. They found no correlation between blood pressure and the final diagnoses. In another study, Dighe et al. [22] reported that the utility of sonographic elastography performed using carotid artery pulsation as a compression source to measure the systolic thyroid stiffness index has the potential to substantially reduce the number of fine-needle aspiration biopsies by detecting benign nodules. Although carotid artery pulsation has been used in those studies as the compression source for thyroid elastography, it has also been reported that arterial pulsations may generate compressiondecompression movements that may create interfering elastographic images with unnecessary thyroid movement and it is difficult to restrict thyroid movement [15]. Hence, in the present study, carotid artery pulsation was not used as a compression source.
Microcalcifications are also characteristic findings of malignant nodules [15]. Mixed calcifications are defined by the presence of microand macrocalcifications, and some recent studies have reported that the presence of mixed calcifications suggests an increased potential for malignancy [15]. In addition, calcification is a highly concordant finding in the evaluation by radiologists [15]. The MicroPure imaging algorithm (Toshiba) is a recently developed imaging technique that aids in the detection of the calcifications by enhancing the superficial structures.
Kurita et al. [23] evaluated the usefulness of MicroPure imaging in measuring the microcalcifications in breast lesions. They reported that MicroPure imaging improved the visualization of microcalcifications and suggested that this imaging algorithm is a clinically useful easy imaging technique in the diagnoses of microcalcifications.
Sankaye et al. [24] examined 25 women through breast sonography, and 11 breast malignancies were diagnosed. Four (16%) (three malignant and one benign) calcifications were visualized through ultrasound. All calcifications were detectable using both B-mode and MicroPure imaging. The authors suggested that all four calcifications appeared more subjectively conspicuous using MicroPure imaging than in the B-mode imaging. In our study, 17 (26.6%) of 65 benign thyroid nodules and three (33.3%) of nine malignant thyroid nodules showed microcalcifications by MicroPure imaging.
Fine-needle aspiration biopsy of thyroid nodule is widely accepted as the most accurate, sensitive, specific, and cost-effective diagnostic procedure in the preoperative assessment of thyroid nodules, with low rates of false-positive (2.3%), false-negative (0.2%), and inadequate results [6, 7]. The accuracy of the fine-needle aspiration biopsy analysis approaches 95% in the differentiation of the benign nodules from the malignant nodules of the thyroid gland [6–8]. The use of ultrasound guidance improves the diagnostic yield [8]. Organizations such as the American Thyroid Association and the American Association of Clinical Endocrinologists suggest that the original cytologic diagnosis can be accepted until the nodule grows or changes in appearance [8]. Instead of routine repeated fine-needle aspiration biopsy, several researchers recommended the use of both repeated fine-needle aspiration biopsy and clinical follow-up together [8]. Because fine-needle aspiration biopsy is used as the reference standard in the present study, limitations of fine-needle aspiration biopsy of thyroid nodule also become limitations of this study. In this study, to eliminate false-positive results, nine patients with cytologic malignancies underwent surgery, and their pathologic diagnoses were confirmed. Also, eight cytologically benign nodules were surgically removed and their pathologic diagnoses were confirmed. To reduce the false-negative results, 57 cytologic benign nodules were monitored clinically and by ultrasound for 12 months, and repeat biopsies were performed.
There are some limitations to this study. First, nodules larger than 40 mm are difficult to evaluate accurately because of the difficulties in measuring thyroid nodule elasticity. Practically, a large size can be a limitation in the nodule-to-gland or nodule-to-muscle strain index described in literature [19]. However, in our study, no single nodule was larger than 40 mm. Pure cystic anechoic nodules without solid components and shell-calcified nodules may cause some measurement problems as well, because of posterior shadow or posterior enhancement artifacts of ultrasound imaging. However, we avoided pure cystic, anechoic, and shell-calcified nodules. During sonographic elastography of the thyroid, it is important to maintain a light pressure on the probe because strong pressure may lead to a misdiagnosis [15]. In addition, carotid pulsation is another restriction in the strain index that may have a negative effect on our study; however, the radiologists were experienced in this specific area. MicroPure imaging is a novel and useful method for the detection of especially suspicious microcalcifications, but more experience is needed to evaluate the utility of MicroPure imaging in the diagnoses of microcalcifications.
In conclusion, elastography is a promising imaging technique that can assist in the differential diagnosis of malignant and benign thyroid nodules. The combination of real-time sonographic elastography, MicroPure imaging techniques, and B-mode sonography may be helpful for the improvement of the differential diagnosis of thyroid malignancies.
Received February 26, 2011.
Revision received July 20, 2011.
© American Roentgen Ray Society
