Abstract
Background/aim: Previous studies have shown that there may be a diversity in the ultrasonographic (US) features discriminating a malignant from a benign thyroid nodule. We determined the reliability of the specific nodule shape in combination with other US features in predicting thyroid carcinomas. Patients and Methods: This was a retrospective single-center study investigating the association of the morphological characteristics of nodular goiters from preoperative US and color Doppler images with malignancy based on pathology. Results: We evaluated 254 thyroid nodules (malignant, n=131) from 205 patients. Multivariable logistic regression analysis showed that taller-than-wide shape [odds ratio (OR)=25.3, 95% confidence interval (CI)=5.4-118.9; p<0.001], microcalcifications (OR=4.9, 95% CI=2.5-9.5; p<0.001), hypoechogenicity (OR=4.5; 95% CI=2-10.3; p<0.001) and size (OR=0.93; 95% CI=0.89-0.98; p=0.002) were independently associated with thyroid nodule malignancy. Additionally, we found a strong negative correlation between size and taller-than-wide shape of a thyroid nodule (r=−0.41, p<0.001). Conclusion: Among the important indicators of thyroid malignancy, taller-than-wide nodules with microcalcifications are most likely to be malignant.
Thyroid nodules are common lesions and are being diagnosed with increasing frequency (1, 2). They are discovered either clinically or using sensitive imaging techniques (3). High-resolution ultrasonography (US) is a powerful tool for detecting thyroid lesions and identify nodules even with a size of a few millimeters (4, 5). As the probability of a first diagnosed nodule being malignant is almost 7% (4), it is important for the clinician to be able to accurately distinguish between benign and malignant nodules in order to make the best clinical decision and minimize unnecessary surgical procedures.
Currently, US-guided fine-needle aspiration biopsy (FNAB) remains the gold standard within the diagnostic procedures in the workup of thyroid nodules. However, a significant number of cases give false-negative, nondiagnostic, indeterminate or suspicious results that are difficult for clinicians to assess (6-8). In this way, US and color Doppler standardized criteria are good diagnostic tools indicating position, size, margins, content, echogenicity, elasticity and vascular pattern of the nodule. For example, marked hypoechogenicity, microcalcifications, irregular margins, intranodular vascular images and low elasticity have been associated with malignancy (4, 9-13). Studies have recently shown that a round appearance or a taller-than-wide shape of the nodule can be a US pattern suggestive of malignancy and recently the shape of a nodule has been added to the imaging criteria for suspicious lesions (12, 14).
We aimed to investigate the association of the morphological characteristics of nodular goiters from preoperative US and color Doppler images with malignancy based on pathology.
Patients and Methods
Study design and patient population. This was a retrospective, single-center, cross-sectional study. We included patients with a nodular goiter who underwent routine preoperative US followed by total thyroidectomy between January 2012 and December 2017. The medical records of these patients were reviewed for demographics and thyroid nodule imaging and pathology results. All conventional and power color Doppler US of thyroid nodules were performed using a 6-12 MHz linear array transducer, logic 7 (KPI Healthcare Inc. Yorba Linda, CA, USA). The US findings were re-evaluated and interpreted by two radiologists with more than 20-year experience who were blinded to the final pathology results. The mean length of time between US and thyroidectomy was 2 months. Regarding multinodular lesions, only nodules more than 6 mm were finally evaluated. The investigated US features of the thyroid nodules included taller-than-wide shape (defined as an anteroposterior to transverse diameter ratio ≥1 in the axial plane), hypoechogenicity, microcalcifications (defined as ≤1 mm), irregular margins and intranodular vascularity. The study was approved by the Institutional Scientific Committee (00/16.10.2017).
Statistical analysis. Descriptive statistics are reported as the median and interquartile range (IQR) for continuous variables, and frequency and percentage for categorical variables. Continuous variables were compared using the Mann-Whitney U-test, while discrete variables were compared using the chi-square or Fisher's exact test, as appropriate. Univariate and multivariable logistic regression analyses were performed to determine the independent effects of sonographic variables associated with thyroid nodule malignancy using the Wald backward-selection method. Only statistically significant variables (p<0.05) from univariable analysis were entered the multivariable analysis. A receiver operating characteristic curve analysis was performed to assess the diagnostic performance of the US features distinguishing benign from malignant thyroid nodules. Correlation between nodule size and taller-than-wide shape was also analyzed by bivariate linear regression analysis. A p-value of less than 0.05 was considered as statistically significant. All analyses were performed using SPSS version 23.0 (IBM, Armonk, NY, USA).
Results
We evaluated 254 nodules from 205 patients (female, n=176) (Table I). Regarding the benign nodules (n=123) the diagnosis included nodular goiter (n=107), adenoma (n=5) and Hashimoto's nodule (n=11), while the great majority of malignant nodules were papillary carcinomas (n=129) and only two were follicular carcinomas.
Univariate and multivariate logistic regression analyses are shown in Table II. Size [odds ratio (OR)=0.93, 95% confidence interval (CI)=0.89-0.98; p=0.002], taller-than-wide shape (OR=25.3; 95% CI=5.4-118.9; p<0.001), microcalcifications (OR=4.9, 95% CI=2.5-9.5; p<0.001), hypoechogenicity (OR=4.5, 95% CI=2-10.3; p<0.001) of thyroid nodules were the only variables independently associated with thyroid malignancy (Table II). Taller-than-wide shape compared to the other US features demonstrated the highest specificity, at 98%, for distinguishing malignant from benign thyroid nodules (Table III), reaching 100% when combined with the presence of microcalcifications (data not shown). Moreover, we found a strong negative linear correlation between the taller-than-wide shape and the size of the nodules, r=−0.41, p<0.001 (Figure 1).
Discussion
This study indicates that in a clinical sample of patients with nodular goiters referred to our center, the majority of nodules with papillary carcinomas had a taller-than-wide shape, significantly higher than that of benign. Additionally, a significant number of carcinomas had micro-calcifications, but the most important finding was the coexistence of micro-calcifications in nodules with a taller-than-wide shape which represented the one third of our malignant nodules. None of the benign nodules with taller-than-wide shape had microcalcifications, increasing the specificity to 100%.
From the results of multiple studies is suggested that none individual US feature can accurately discriminate a malignant nodule and US alone cannot be used for the decision of surgical intervention (15-17). Furthermore, the Thyroid Imaging Reporting and Data System, which is applied primarily to exclude a thyroid malignant lesion, until now is not widely accepted to allow the clinicians for a proper therapeutic decision (18). Again, many studies evaluated the US findings by comparison with the cytological results after FNAB (16-18) but several factors can affect the diagnostic accuracy of FNAB, such as the experience of physicians in its performance, sampling error, non-diagnostic reports or cytological atypia of undetermined significance (19).
In several studies, the main suspicious features indicating thyroid malignancy were found to have different diagnostic power among different cohorts, as demonstrated from the ORs in multivariate analysis (17, 20, 21). For example, Liu et al. (20) and Moon et al. (21) found that microcalcifications were a stronger predictor of papillary carcinoma among other US features, following by nodule shape. Additionally, Capelli et al. reported an OR for malignant nodules of 8.6 for those with taller-than-wide shape (17).
In the current work, hypoechoic nodules, nodules with ill-defined margins, or those with intranodular vascularity had a high OR for malignancy but the highest OR was found for nodules with macrocalcifications and those nodules with taller-than-wide shape. In our cases, taller-than-wide shape, hypoechogenecity and microcalcifications were the most important factors predicting malignancy in multivariate regression analysis. Taller-than-wide nodule shape had a much higher OR than microcalcifications (33.8 vs. 6.7), whereas intranodular vascularity or ill-defined margins were not enough to determine a high risk of malignancy, as previously described (22). This was confirmed by the AUCs which are also used to evaluate the diagnostic accuracy. The AUCs of shape, microcalcifications and the combination of both were 0.671, 0.721 and 0.649 respectively, demonstrating relatively good accuracy, showing the combination also to be as useful as the shape or the presence of microcalcifications alone in predicting malignancy. It is known that the coexistence of two or more highly suspicious US criteria greatly increases the potential risk for nodules being malignant (17, 23). In our work, the combination of microcalcifications and taller-than-wide shape was an independent predictor of malignancy and although the sensitivity was only approximately 30%, the specificity was 100%, since none of the benign nodules with taller-than-wide shape were found to have microcalcifications. This means that the coexistence of microcalcifications in taller-than-wide nodules provides strong evidence of malignancy, although the absence of this profile cannot exclude malignancy.
It should also be pointed out that marked hypoechogenecity was common in our malignant nodules, with an OR of 6.5, higher than previously reported (24) indicating an increased risk of malignancy, whereas intranodular vascularity or ill-defined margins were adequate for determining a high risk of malignancy. When hypoechogenecity was evaluated in combination with microcalcifications and with the shape, our results showed no improvement of sensitivity or higher AUC (data not shown). In our study, nodules with taller-than-wide shape or shape combined with microcalcifications were the best predictors of malignancy.
It should also be noted that the size of malignant nodules was significantly lower than that of benign ones, reflecting the strong negative correlation between the nodule size and taller-than-wide shape. This might mean that smaller nodules with taller-than-wide shape are more often malignant than larger ones. Previous studies have examined the correlation between malignancy and nodule size (25, 26). These studies included large nodules (>3 cm) and size correlated with the FNAB results and other US findings but not with the shape as evaluated in our study. Hammad et al. proposed a cut-off point of 6 cm where the incidence of carcinoma decreases with increasing size of thyroid nodules (26). The anteroposterior nodule diameter might be suitable for assessing thyroid malignancy in microcarcinomas, with a critical value >0.7 cm (12).
In our study, one limitation included a lack of a large number of nodules more than 4 cm in size. The explanation for this is that clinicians often recommend thyroid US to their patients and thyroid cancer is increasingly discovered at an early stage (27). A strength of our study is that there were final pathological results for all nodules after thyroidectomy and there was no selection bias since all patients who had undergone operations were included, without exception, and this explains how the potential malignancy rate was higher in comparison with the non-operated on nodules. On the other hand, the main limitation of the study was the relatively small number of nodules included and that it was a single-center study.
In conclusion, the current study emphasizes that among the important indicators of thyroid malignancy, taller-than-wide nodules with microcalcifications are most likely to be malignant and surgical treatment should be considered for these patients. The strong negative correlation between thyroid nodule size and the taller-than-wide shape (potential malignancy) needs additional studies including more larger nodules.
Footnotes
Authors' Contributions
K.D.P: study concept and design, data analysis and interpretation, statistical analysis and article writing; H.J.K: study concept design, data analysis and interpretation and critical revision of the article; C.C.E., I.A.I., K.S.M., N.I.K.: data acquisition and interpretation and critical revision of the article; G.N.K.: statistical analysis and critical revision of the article.
This article is freely accessible online.
Conflicts of Interest
The Authors declare no conflicts of interest in regard to this study.
- Received March 14, 2020.
- Revision received April 3, 2020.
- Accepted April 8, 2020.
- Copyright© 2020, International Institute of Anticancer Research (Dr. George J. Delinasios), All rights reserved