Correlation of the Thyroid Nodules' Sonographic Features With Fine Needle Aspiration (FNA) Cytology Results

Background: Thyroid nodules are a common nding in clinical practice. Although ultrasonography is an accepted method for evaluating and following these nodules, Fine Needle Aspiration (FNA) is the procedure of choice for assessing the risk of malignancy. This study aims to determine the correlation between sonographic features of thyroid nodules based on Thyroid Imaging Reporting and Data System (TIRADS) classication and the cytology results obtained by FNA of thyroid nodules. Methods: In this prospective cohort study, 147 patients with thyroid nodules underwent FNA under the guide of ultrasonography based on TIRADS classication, and their sonographic features were recorded. The pathologic ndings were also obtained according to the Bethesda system. Finally, the association between sonographic features and cytology results were analyzed. Results: 147 patients with a mean ± SD age of 49.8 ± 13.7 years were assessed. 16 (10.9%) nodules were malignant, and 131 nodules (89.1%) were benign. The association of TIRADS categories with the risk of malignancy is as follows: TIRADS 1 (0%), TIRADS 2 (16.9%), TIRADS 3 (10.5%), TIRADS 4 (16.7%), and TIRADS 5 (0%). The location of thyroid nodules and their bloody lamellae were signicantly correlated with the risk of malignancy (P value< 0.05). However, the association between the risk of and gender, calcication, hardness, halo sign and nodules’ echogenicity were not statistically signicant. Conclusions: Although there are trusted classications for categorizing the thyroid nodules, there is still uncertainty in utilizing them as an accepted method of choice for managing the thyroid nodules, as various sonographic features are shared between benign and malignant ones.

Conclusions: Although there are trusted classi cations for categorizing the thyroid nodules, there is still uncertainty in utilizing them as an accepted method of choice for managing the thyroid nodules, as various sonographic features are shared between benign and malignant ones.

Background
In clinical practice, thyroid nodule is a frequent nding, which is revealed during a precise physical exam or a variety of imaging procedures [1]. According to epidemiological documents, in iodine su cient regions of the world, near 5% of the females and 1% of the males are detected with palpable thyroid nodules [2]. However, they can be discovered in 10-41% of adults by ultrasonography (US), which is increasing with recently growing use of this imaging procedure [3] and it is stated that most of these nodules are benign [4]. Although there is an increasing incidence of thyroid nodules due to improved health care access of population [5], there is still an urge to increase the rate of diagnosis at the lowest cost and least possible time.
Ultrasonography (US) is one of the most common rst-line modality of evaluating palpable thyroid nodules or detecting them incidentally [6]. However, most of these nodules should undergo Fine Needle Aspiration (FNA), which is the keystone of assessing nodular thyroid disease, to increase the accuracy of predicting their potential cancer risk [7,8]. FNA under the guide of US has made biopsy the most important tool for making decision about conservative management or surgery of thyroid nodules [9].
There has also been a substantial controversy over assessing clinically asymptomatic nodules by FNA biopsy or close observation [10].
Thyroid Imaging Reporting and Data System (TIRADS) classi cation, which was rst established in 2009, is a model for evaluating the thyroid nodules based on speci c patterns of US that enhances the selection of nodules for further evaluation by FNA biopsy [11]. Although there are some sonographic features which are suggestive of malignancy of the thyroid nodules, there is extreme variability in reported sensitivity and speci city of these ndings in correlation with nal cytology results from study to study [3].
The aim of our study was to compare the cytology reports of thyroid nodules' FNA biopsy with their speci c sonographic features in predicting the risk of thyroid malignancy. In other words, we tried to determine US characteristics which have the most correlation with malignant tumors to reduce the number of unnecessary FNA procedures in patients with thyroid nodules and their suitable selection.

Methods
In this prospective cohort study, a total of 147 consecutive patients with thyroid nodules, who referred for FNA biopsy due to any reason, underwent cytological study and US after obtaining informed consent at an institution from December 2018 to September 2019. Our inclusion criteria were based on any patient who referred for further evaluation of his/her suspicious thyroid nodules due to any reason. Those who had a history of benign cytology result and previous thyroid malignancy were excluded from the study.
Neck US was performed for each patient by our experienced radiologist using a high-resolution US apparatus (Medison Accuvix V10, Korea) with 7 MHz linear transducer. Every thyroid nodule was assessed based on ACR TI-RADS criteria such as composition, echogenicity, shape, margin, and echogenic foci [12]. The TIRADS model categorizes the nodules into 5 groups ranging from 1 (benign) to 5 (highly suspicious), as summarized in Table 1. It is noteworthy that all 147 patients underwent FNA, regardless of the nodules' TIRADS grade, to determine the accuracy of ACR-TIRADS classi cation system. clinician underwent FNA biopsy under the guide of US, which was performed for them in supine position with a mid-extended neck position under standard sterilized condition by our quali ed interventional radiologist. Lidocaine was used as the local anesthesia. Target nodules were aspirated with a 21-gauge needle attached to a 20 cc syringe, using aspiration technique; the obtained samples were xed with alcohol on glass slides. Our pathologist reported the cytology based on Bethesda system.
Finally, the cytology results and TIRAD scoring of these nodules were recorded and analyzed using SPSS software. The statistical tests used included Chi-square and Fisher and Kappa for comparison of categorical variables. The P-value less than 0.05 was considered statistically signi cant. In addition, the sensitivity, speci city, positive predictive value, and negative predictive value of TIRADS were determined.
It is noteworthy that all 147 patients underwent FNA regardless of the nodules' TIRADS grade to determine the accuracy of ACR-TIRADS classi cation system.

Results
Totally, 147 nodules of 147 patients were assessed. The patients' mean age was 49.8 years ± 13.7 (standard deviation) and 126 patients were female (85.7%). The age ranges of patients were from 20 to 80 years.

Discussion
Fine Needle Aspiration Biopsy (FNAB) is accepted as a precise, safe, and cost-effective tool for diagnosis of the thyroid nodules, especially for preoperative decision making [13], and Sonography guided FNAB has also improved its diagnostic accuracy remarkably recently [14]. Most of the time, the purposeful or incidental nding of a thyroid nodule is rst diagnosed by Ultrasonography (US) [15]. Therefore, physicians usually decide if the nodule needs further evaluation based on these nodules' sonographic features. However, most of these sonographic characteristics are shared between benign and malignant lesions [16]. In this study, we tried to compare these sonographic features based on the ACR TIRADS model with nal cytology reports of the thyroid nodules and evaluate how these features could make a physician suspicious of malignancy and additional procedures.
Despite an increased rate of thyroid nodule diagnosis following a growing variety of modalities, most of the nodules are benign [17]. As in our study, 89.1% of the nodules were benign lesions, including benign follicular nodule, cystic uid only, Colloid nodules, etc. However, different factors affect this prevalence, such as iodine de ciency, post-radiation therapy, and even patients' level of health service accessibility in different societies [17,18].
According to our experience, like some other studies [19,20], the size of the nodules is not a good indicator of the risk of malignancy of thyroid nodules. However, some researches claim that there is a strong association between these two factors as the larger size of the nodules correlates with a higher malignancy rate with a threshold of 2.0 cm [21,22]. Generally, the nodules' size is not an important predictive indicator of thyroid malignancy unless accompanied by other malignant features [19].
In this study, surprisingly, we found no signi cant correlation between the TIRADS model and the risk of malignancy (p value = 0.45); this is in contrast with what Singaporewalla et al. [23] claimed in their research. They showed that there was an accuracy of 83% in predicting the risk of malignancy based on using TIRADS. However, the maximum correlation that we found was in TIRADS 2, with 16.9% of malignancy association. There are other studies by Horvath et al. [11], Park et al. [24], and Kwak et al. [25] that reported the leading association of TIRADS with the risk of malignancy 89.6%, 100%, and 87.5%, respectively. This may be due to our limited sample size, but notably, the TIRADS model is somehow operator dependent, as some studies suggest using this model by two radiologists at the same institution.
As mentioned above, all 147 patients underwent FNA under the guide of sonography to determine the accuracy of ACR-TIRADS classi cation and the correlation of histologic ndings in accordance with sonographic features. In our experience, interestingly, TIRADS 2 had the most correlation with the risk of malignancy, which increases the uncertainty of using this scoring system as an indicator of thyroid nodules' FNA necessity. We believe that this discrepancy may cause the misdiagnosis of patients who present with thyroid nodules with TIRADS 2 and will not be further assessed according to ACR-TIRADS.
Our study showed that there was a statistically signi cant (p = 0.013) correlation between malignancy and the location of thyroid nodules, and most of our cytologically malignant nodules were located in the right lobe. Although 67 out of all 77 nodules of the right lobe were benign, 10(13%) were malignant compared to 5 (7.7%) out of 65 nodules of the left thyroid lobe. Most studies claim that there is no relationship between the location of the thyroid nodules and malignancy and despite the absence of association, the isthmus and mid-lobar nodules were the most sites of the thyroid, which correlates with the risk of malignancy [26,27].
In accordance with a retrospective observational cohort study conducted by Frates et al. [28], one of the sonographic characteristics which correlate with malignancy would be the existence of microcalci cation. This nding is also in contrast with our study in which there was no association between the nature of calci cation and the risk of malignancy (p value = 0.9). In another study by Rago et al. [29], only a combination of the presence of microcalci cation plus the absence of halo sign had a signi cant relationship with the possibility of malignancy, which is associated with high speci city (93.0%) but low sensitivity (36.0%). Our research did not differentiate the subtypes of calci cations (such as micro, macro, coarse, and peripheral) and this may affect the results as some studies showed that although malignant thyroid nodules might correlate with microcalci cations, benign lesions are even associated with macro-calci cation [30,31].
Several studies have tried to determine if Fine Needle Capillary (FNC) sampling is superior to Fine Needle Aspiration due to a higher amount of cellular material, but they did not obtain statistically signi cant data [32,33]. Although FNA sampling provides a more speci c eld and more diagnostic parameters than the FNC technique [33], our experience revealed that the samples accompanied by blood were signi cantly associated with more risk of malignancy (p value = 0.047).
In this study, we evaluated a relatively new variable to determine if the angle of the needle entry to thyroid nodules during FNA sampling affects the risk of malignancy. According to the results, there was no signi cant (p value = 0.47) correlation between the group of patients in whom the needle was induced in the angle of 75º and the other group who underwent this procedure with any other angle of needle entry.
Based on our experience in this research, hypoechogenicity is somehow associated with malignancy with a p-value of 0.058, which was statistically insigni cant. However, Nabahati et al. revealed that there was a considerable positive correlation between malignancy and hypoechogenicity [odds ratio (OR) 3.577, 95% con dence interval (CI) 2.045-6.256] as their 29 nodules out of all 221 hypoechoic nodules were malignant (p value < 0.001). In our study, 26% of the thyroid nodules were hypoechogenic, and 20% were malignant. Overall, echogenicity is one of the criteria of ACR TI-RADS, and as in our study, the more hypoechogenic the thyroid nodule, the higher the risk of malignancy.

Conclusions
Thyroid nodules are a common nding of clinical practice worldwide. The clinical approach to this clinical nding is based on the rst sonographic features and the FNA biopsy results. Although there are accepted models such as TIRADS to distinguish malignant lesions with high sensitivity and speci city, our study demonstrates that even these relatively trusted scoring systems may not be trustworthy enough and most of these sonographic features are mutual between malignant and benign lesions. Declarations Acknowledgements: The authors would like to thank Shiraz University of Medical Sciences, Shiraz, Iran and also Center for Development of Clinical Research of Nemazee Hospital and Dr. Nasrin Shokrpour for editorial assistance.
Authors' contributions HG, AAr designed the study and obtained ethical approval, MA collected the clinical data and HH carried out the data gathering. AAr and HH drafted the manuscript while HG and AAb edited and prepared the nal version of the article. All author proofread and approved the nal version of the manuscript.

Funding
None.
Availability of data and materials The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Ethics approval and consent to participate
The study protocol was approved by the medical Ethics Committee of Tehran University of Medical Sciences (Ethics ID: IR.TUMS.VCR.REC.1398.686)

Consent for publication
Consent was obtained from the patient parent/guardian regarding the publication of this case report.