Role of sonographic features in predicting ……….Al-Ass Role of sonographic features in predicting the malignant potential of thyroid nodules

However, while thyroid nodules are common, thyroid malignancy is relatively rare, constituting about 1% of all malignancies. The main point of the approach of the thyroid nodule is the detection of the malignant nodules and deciding for the surgical treatment. Ultrasonography (US) plays a crucial role in the diagnostic management of thyroid nodules. The aim was to assess the reliability of using hypoechoic, solid and ill-defined margin as independent predictors for the identification of malignant thyroid nodules on US. We retrospectively analysed the three suspicious US features of malignancy for 145 patients with 255 thyroid nodules who underwent thyroid resection. We used histological results as gold standard reference test. Of the 255 surgical resected nodules; hypoechoic nodules had a sensitivity of 66.7%, and positive predictive value (PPV) of 8.7%. Solid nodules had a sensitivity of 33.3% and PPV of 3.7%. Ill-defined margins nodules had a sensitivity of 66.7% and PPV of 6.9% in predicting malignancy. The present study adds further evidence on the poor PPV in our results, indicated, that individual US features are not reliable used as independent predictors for the identification of malignant potential thyroid nodules on US. malignant nodules were ill-defined margins, compared with 93.1% of benign nodules. Ill-defined margins nodules had a sensitivity of 66.7%, specificity of 43.8% and PPV of 6.9% in predicting malignancy. In the literature, the ill-defined margins nodules had various values ranging from 17.4% to 77.5% for sensitivity, 38.9 – 85% for specificity and 9.3 – 60% for PPV [1, 4, 7, 9, 12, 13, 20, 22, 24]. Our results agree with the currently data in the literature for sensitivity and specificity. In our study, a hypoechoic, ill-defined margins and solid nodules had very low PPV indicating that many of the positive results from this testing procedure are false positives. These findings make a hypoechoic, solid and ill-defined margins as a poor reliable and single predictors of malignancy in a thyroid nodule on US in our study. Our findings support previous study results 4, 5, 8, 14, 25]. Thus it will be necessary to follow up any positive result with a more reliable test (Histopathology) to obtain a more accurate assessment as to whether cancer is presented. The reveals great variability in the findings of studies. of all, US is an observer – dependent method. Interobserver variability are reported, particularly on descriptions of echogenicity and margins features of thyroid nodules the can be explained by the they one US of is categorized as malignant nodule. to the differing methodologies.

US is an important diagnostic tool in predicting thyroid malignancy and selecting thyroid nodules that should be evaluated by fine needle aspiration cytology (FNAC) [2, 3, 22 -24]. Thus, preoperative neck US has a role in surgical planning [23]. It is safe, non-invasive and cost effective diagnostic tool for preoperative assessment of patients with thyroid nodules to help the surgeon in the management of these nodules [1,4,8,13,14,16,25]. The purpose of our study was to evaluate the reliability of using the US features, such as hypoechoic, solid and ill-defined margin, in predicting thyroid malignancy in our clinical practice.

Methods
This study is a retrospective carried out at three main hospitals; Al-Gamhouria Modern General Hospital, Basuheeb Military General Hospital and 22 May Hospital in Aden city, Yemen, for two years (2014 and 2016). We reviewed the three suspicious US features of malignancy for 255 nodules in 145 patients. We wanted to test the hypothesis that preoperative these three US features were independent predictors of malignancy.
The criteria for these three US features that we used in this study are based on previous studies [3, 5, 13, 14, 23 -25]. Thyroid nodule is defined as a discrete lesion within the normal thyroid parenchyma [3,5,23,24]. Hypoechoic nodule is defined as lower echogenicity (darker) when compared to the surrounded thyroid parenchyma or the adjacent strap muscles [3,13,14,24,25]. Solid nodule is defined as no obvious cystic component or cystic component accounting for ≤ 10% of the nodule volume [3, 5, 24]. Ill-defined margin nodule as is defined as poorly demarcated margin which cannot be obviously differentiated from adjacent thyroid tissue [3,5,13,24]. According to these definitions, sonographically, the nodules were assessed on the basis of hypoechoic, solid and ill-defined margin US features were considered malignant. Nodule size was recorded as the largest of the three dimensions: length, width, and depth. We looked in each report for these three features that matched our definitions.
All patients underwent surgery for the suspicion of malignancy; information on patient demographics, US characteristics, FNAC and final histology of the nodules were collated. All the cases of thyroiditis were excluded. We considered histopathological diagnosis of resected thyroid gland tissue after surgery to be the gold standard reference test.

Statistical Analysis
The US diagnosis was compared with cytology reports and the histological diagnosis. The sensitivity, specificity, PPV and diagnostic accuracy of US were calculated based on combination of those three US features that were exhibited in 115 thyroid nodules. Each US feature was also compared with its pathology results. Outcomes of interest were the sensitivity and PPV of each US feature of thyroid nodules. The sensitivity, specificity and PPV for each feature were calculated independently in predicting malignancy.
The data were analysed using SPSS version 17.0 (SPSS Inc., Chicago, IL, USA). Baseline continuous data (age and tumor diameter [nodule size]) were expressed in mean ± standard deviation (SD) with confidence interval (95% CI), and categorical data were expressed as percentages. Univariate associations between the presence of malignancy and discrete variables were evaluated using Pearson Chi -Square test (χ 2 ). Independent sample Student's ttest was used to compare between categorical and continuous variables. P ≤ 0.05 was considered statistically significant.
In our study, US was able to correctly identify 10 out of 15 malignancies, and 135 out of 240 benign nodules. The diagnostic yield of US for malignant thyroid nodule, including sensitivity, specificity, PPV and accuracy, were 66.7%, 56.3%, 8.7% and 56.9% respectively. According to the literature, the diagnostic yield of US has different values ranging from 20% to 96% for sensitivity, 59% -100% for specificity, 15 -100% for PPV and 59 -94% for accuracy [4 -6, 11, 13, 16, 26, 27]. Our results concur with currently data in the literature for sensitivity and nearly similar to data in the literature for specificity and accuracy.
In evaluating thyroid nodules, we analysed three individual US characteristics of nodules in predicting malignancy included hypoechogenecity, solid component and ill-defined margins. We compared these characteristics to their respective pathology.
Although most malignancies tend to be hypoechoic, benign nodules may be hypoechoic too [21]. Therefore, many thyroid cancers would be missed if only the hypoechoic nodules underwent FNAC [2, 3]. In our study, 8.7% of malignant nodules were hypoechoic, compared with 91.3% of benign nodules. Hypoechoic nodules had a sensitivity of 66.7%, specificity of 56.3% and PPV of 8.7% in predicting malignancy. In the literature, the hypoechoic nodules had various values ranging from 26.5% to 87.1% for sensitivity, 43.4 -94.3% for specificity and 11.4 -68.4% for PPV [1,4,9,11,13,14,20,24]. Our findings agree with the currently data in the literature for sensitivity and specificity. In contrast to the literature, our results showed lower PPV.
Thyroid malignant nodules are most commonly solid or nearly entirely solid and are more likely to be solid than benign nodules [10, 14, 24]. In our study, 3.7% of malignant nodules were solid, compared with 96.3% of benign nodules. Solid nodules had a sensitivity of 33.3%, specificity of 45.8% and PPV of 3.7% in predicting malignancy. In the literature, the solid nodules had various values ranging from 69% to 75% for sensitivity, 52.5 -55.9% for specificity and 15.6 -27% for PPV [1,4,8,9,14,24,25]. In contrast to the literature, our results showed lower values for sensitivity, specificity and PPV respectively.
When more than 50% of the margin of a thyroid nodule is not clearly defined, it is considered as poorly defined [3,24]. In fact, malignant nodules tend to have ill-defined margins due to the infiltration of the surrounding thyroidal parenchyma [23,24]. Therefore, irregular margins are finding highly suggestive of malignancy [1, 3, 23]. Unfortunately, this finding is also reported in benign conditions such as thyroiditis [4,8,14,22,25] or in some benign thyroid nodules incompletely encapsulated that can merge with normal tissue [23,24]. In our study, 6.9% of Univ. Aden J. Nat. and Appl. Sc. Vol. 23 No.2 -October 2019 malignant nodules were ill-defined margins, compared with 93.1% of benign nodules. Ill-defined margins nodules had a sensitivity of 66.7%, specificity of 43.8% and PPV of 6.9% in predicting malignancy. In the literature, the ill-defined margins nodules had various values ranging from 17.4% to 77.5% for sensitivity, 38.9 -85% for specificity and 9.3 -60% for PPV [1,4,7,9,12,13,20,22,24]. Our results agree with the currently data in the literature for sensitivity and specificity.
In our study, a hypoechoic, ill-defined margins and solid nodules had very low PPV indicating that many of the positive results from this testing procedure are false positives. These findings make a hypoechoic, solid and ill-defined margins as a poor reliable and single predictors of malignancy in a thyroid nodule on US in our study. Our findings support previous study results [1,4,5,8,14,25]. Thus it will be necessary to follow up any positive result with a more reliable test (Histopathology) to obtain a more accurate assessment as to whether cancer is presented.
The literature reveals great variability in the findings of studies. First of all, US is an observerdependent method. Interobserver variability are reported, particularly on descriptions of echogenicity and margins features of thyroid nodules [3]. Second, the difference can be explained by the fact that they considered presence of even one US criterion suggestive of malignancy is categorized as malignant nodule. Third, likely due to the differing methodologies.
Although US features utility in predicting malignancy has caused profound changes in the management of thyroid nodules. Finally, in the literature , there is consensus on no single US feature or combinations of features are adequately sensitive to identify all malignant nodules [7 -17, 19 -28]. Thus, the goal of management should be to avoid extensive and costly evaluations in most patients with benign disease without missing the minority of patients who have thyroid cancer.
To our knowledge, our study is the first to report on three specific US characteristics of thyroid nodule malignancy that have been published to date in Aden -Yemen. The strength of our study was that the US examinations were performed by the same radiologist, which reduced diagnostic variability. All the results of US features in this study were compared with the gold standard criterion (histology). Our study is primarily limited by its retrospective nature.

Conclusions
In spite of thyroid US can be helpful in the differentiation of benign from malignant lesions. The fact that no US features were pathognomonic for malignancy. In this study, we have found that a hypoechoic and ill-defined margins nodules had moderate sensitivity while solid nodules had low sensitivity. All these features had very low PPV in predicting malignancy on US. Thus, it should not be used as a reliable sole predictors for the identification of malignant potential thyroid nodules in our clinical practice. The diagnosis of thyroid cancer is a major obstacle that needs to be overcome in the future. We emphasized that careful analysis of any features is imperative to identify the thyroid cancer.