Impact of thyroid nodule sizes on the diagnostic performance of Korean thyroid imaging reporting and data system and contrast-enhanced ultrasound

OBJECTIVE: This study aimed to evaluate the impact of thyroid nodule sizes on the diagnostic performance of Korean thyroid imaging reporting and data system (TIRADS) and contrast-enhanced ultrasound (CEUS). METHODS: In total, 308 consecutive patients with 382 thyroid nodules underwent US-guided FNA or surgery were included in this retrospective study. The nodule size was classified into 3 categories: ≤10 mm (group A), 10–20 mm (group B), and ≥20 mm (group C). We compared the risk of malignancy in each subgroup, categorized according to the TIRADS and CEUS patterns. RESULTS: In group A, the differences in diagnostic value between TIRADS and CEUS were significant (AUC: 0.804 vs 0.733, P = 0.028, sensitivity: 81.8% vs 72.7%, P = 0.013, specificity: 88.9% vs 79.4%, P = 0.011). In group B, the AUC (0.897), sensitivity (88.1%) and specificity (91.9%) of CEUS were highest. In group C, the specificity of CEUS was significantly higher compared with TIRADS classification (90.8% vs 82.9%, P = 0.023), while the sensitivity and AUC showed no significant difference between the two models (84.2% vs 81.5%, P > 0.406, 0.848 vs 0.820, P = 0.545). CONCLUSIONS: Nodule size influences the diagnostic accuracy of the two methods. TIRADS have best value in nodules ≤10 mm, while CEUS perform best for differentiating lesions >10 mm, especially in lesions ≥20 mm.

Keywords: Thyroid ultrasound; nodule size; thyroid malignancy; contrast-enhanced ultrasound; TIRADS; Biopsy

The primary size of thyroid nodule may determine an individual's risk for developing thyroid cancer, lymph node spread, and distant metastasis [[1]]. Therefore, Fine-needle aspiration (FNA) is a necessary procedure for many patients with large nodules beyond which cancer risk is higher. However, FNA is imperfect because 20% – 25% of samples return indeterminate, such as cytologic findings reported as "suspicious for follicular neoplasm," however, are relatively nonspecific reflecting an increased risk of malignancy; the overall rate of malignant results is approximately 20% [[2]]. In addition, it remains controversial as to whether a large nodule size has a higher malignancy risk than smaller nodules [[4]]. Sophia et al. reported that increasing thyroid nodule size impacts cancer risk in a nonlinear fashion, and a threshold is detected at 2.0 cm, beyond which cancer risk is enlarge. Conversely, other studies reported that smaller thyroid nodules (smaller than about 2.0 cm) are associated with increased probabilities of malignant disease [[5]].

Conventional ultrasonography (US) allows the identification of a wide spectrum of sizes and characteristics that result in difficulty selecting nodules for FNA biopsy [[7]]. During 2009, Horvath et al. firstly proposed the Korean Thyroid Imaging Reporting and Data System (TIRADS), which is a feature of two-dimensional ultrasound based on thyroid nodules was derived from the breast imaging reporting and data system (BI-RADS) [[8]]. As the number of suspicious US features increased, the fitted probability and risk of malignancy also increased [[10]]. A recent meta-analysis including 10,437 thyroid nodules showed that TIRADS has a good sensitivity (0.79) and specificity (0.71) in diagnosis of patients with thyroid nodules [[11]].

Contrast-enhanced ultrasound (CEUS) reveals the microvascularity of tumor tissue, obtaining a spatial and temporal resolution superior to the traditional color and power Doppler techniques [[12]]. Several early studies have implied that, under CEUS, rim-like enhancement and heterogeneous enhancement suggested as a helpful criterion in the diagnosis of malignant nodules with accuracy of 88.4% and 91.6% [[14]]. A recent study reported that the proportion of thyroid nodules with cytologically normal or abnormal among malignancies differed according to nodule size, which might have different US and CEUS features [[18]]. However, the influence of nodule sizes on the performance of TIRADS and CEUS has not been well investigated. In this study, we examine the impact of the thyroid nodule size on the accuracy of TIRADS and CEUS.

This study was approved by the Ethics Committee of The First Affiliated Hospital of Guangxi University of Chinese Medicine (approval number: 2016112). Between May 2014 and August 2017, 308 consecutive patients (86 male, 222 female) with a total of 382 thyroid nodules underwent US-guided FNA or surgery at our hospital were included in this retrospective study. The mean±SD age of the 308 patients was 43.2±11.5 years (range, 20–66 years). The inclusion criteria were: (1) patients had solid or mainly solid thyroid nodules, and (2) records permitting available FNA results, nodule location, and surgical histology. The exclusion criteria were: (1) patients with neck deformity, (2) insufficient/nondiagnostic thyroid cytology, (3) patients were allergic to sulfur hexafluoride microbubbles (SonoVue), and (4) patients were pregnant. Written informed consent was obtained from each participant before the study.

A conventional US scan using a 4–15 MHz liner transducer GE Logiq E9 (GE Healthcare, Chalfont St Giles, United Kingdom) was performed by one of four experienced radiologists (who had 18, 15, 15 and 13 years of experience, respectively) in thyroid imaging. The US focus was placed at the same level during the examination. We assessed the following US features of thyroid nodules: size (the largest diameter), composition (solid or mixed), margin (clear, unclear), echogenicity (hypoechogenic, isoechogenic, or hyperechogenic), aspect ratio (taller-than-wide), and calcification. As the number of suspicious US features increased, the fitted probability and risk of malignancy also increased. Thyroid nodules were categorized into very low-suspicion (TIRADS 2), low-suspicion (TIRADS 3), intermediate-suspicion (TIRADS 4), and high-suspicion (TIRADS 5) according to the TIRADS. TIRADS 2: Pure cysts with or without comet-tail artifacts, anechoic with hyperechoic spots, and grid aspect. TIRADS 3: Hyper or isohyperechoic, partially encapsulated nodule with no suspicious features. TIRADS 4: Solid or mixed hypoechoic nodules with ill-defined borders and a thick capsule, penetrating vessels, with or without calcifications. Hypoechoic, solid nodules with any suspicious US feature. TIRADS 5: Iso or hypoechoic, solid nodules with any suspicious US feature [[8]].

Following conventional US evaluation, CEUS examination was performed, with a 3–9 MHz liner transducer by one of three experienced radiologists (who had 15, 15 and 13 years of experience in performing thyroid CEUS, respectively). SonoVue (2.4 mL, BR1; Bracco SpA, Milan, Italy) 25 mg of lyophilized powder was quickly pushed into the peripheral vein with a probe, followed by injection of normal saline (5 mL), and mixed uniformity. Dynamic contrast images were recorded and stored after SonoVue administration, which allowed successive or frame-by-frame playback to be conducted by one of three experienced radiologists (who had 15, 15 and 13 years of experience in performing thyroid CEUS, respectively). The parameters recorded included the enhancement intensity, internal enhancement and edge enhancement [[15]]. The enhancement degree criteria were divided into hypo-, iso-, hyper-enhancement and rim-like enhancement, which were regarded as indictors for qualitative analysis [[19]]. The malignant thyroid nodules were diagnosed based on inhomogeneous hypo-enhancement in nodule tissues. The benign thyroid nodules were diagnosed based on high enhancement or equal enhancement, and presence of rim-like enhancement in nodule tissues [[20]].

Statistical analyses were performed with SPSS 16.0 software (IBM Corp., Armonk, NY, USA). The nodule size was classified into 3 categories: ≤10 mm (group A), 10–20 mm (group B), and ≥20 mm (group C). The the 2-tailed Chi-square (χ2) test or Fisher's exact test was used to compare the risk of malignancy in each subgroup, categorized according to the US and CEUS patterns. The χ2 test for trend was used to investigate the trend of the malignancy risk as the nodule size increased. The sensitivity, specificity, accuracy, positive prediction value (PPV) and negative prediction value (NPV) of the benign and malignant thyroid nodule diagnoses by each method were compared by χ2 test. A significant difference was defined as a P value of <0.05.

Among the 382 thyroid nodules, 127 were malignant and 255 were benign. When grouped according to size, there were 85 nodules in group A (≤10 mm), 183 nodules in group B (10–20 mm), and 114 nodules in group C (≥20 mm). In group A, 22 were malignant and 63 were benign; in group B, 59 were malignant and 124 were benign; in group C, 38 were malignant and 76 were benign. Of the 119 malignant nodules, there were 116 (97.5%) papillary thyroid carcinoma and 3 (2.5%) follicular thyroid carcinoma. Of the 263 benign nodules, there were 219 (83.3%) nodular goitre, 40 (15.2%) adenomatous goitre, 3 (1.1%) hashimoto's thyroiditis and 1 (0.4%) subacute granulomatous goitre.

Table 1 shows the malignancy risk of nodule size based on TIRADS classification. TIRADS 2, 3, 4 and 5 were 0, 14.6% (25 of 171 nodules), 48.9% (89 of 182 nodules) and 100.0% (13 of 13 nodules), respectively, with significant differences between categories (P < 0.001).

Table 1 The malignancy rates of TIRADS classifications

Table 2 shows the malignancy risk of nodule size based on CEUS enhancement information. The hypo-, iso-, hyper-enhancement and rim-like enhancement mode were 66.7% (110 of 165 nodules), 14.8% (16 of 108 nodules), 5.1% (4 of 78 nodules) and 6.5% (2 of 31 nodules), respectively, with significant differences between categories (P < 0.001).

Table 2 The malignancy rates of CEUS enhancement information

The ROC curves demonstrated that the best cutoff in all three groups were TIRADS 4 and hypo-enhancement, respectively. In group A, the differences in diagnostic value between TIRADS and CEUS were significant (AUC: 0.804 vs 0.733, P = 0.028, sensitivity: 81.8% vs 72.7%, P = 0.013, specificity: 88.9% vs 79.4%, P = 0.011). In group B, there were no significant difference between the two models (AUC: 0.862 vs 0.897, P = 0.253, sensitivity: 84.7% vs 88.1%, P = 0.289, specificity: 88.7% vs 91.9%, P = 0.352). In group C, the specificity of CEUS was significantly higher compared with TIRADS classification (90.8% vs 82.9%, P = 0.023), while the sensitivity and AUC showed no significant difference between the two models (84.2% vs 81.5%, P > 0.406, 0.848 vs 0.820, P = 0.545) (Table 3).

Table 3 Diagnostic value of TIRADS classifications and CEUS enhancement information for malignant nodules according to nodular size

Our study evaluated the impact of thyroid nodule sizes on the differential diagnosis of benign and malignant of CEUS and TIRADS, and found that the impact of nodule size on the accuracy of diagnostic performance differed according to each method. The TIRADS recommends in its risk stratification chart that FNA be performed on thyroid nodules at a threshold of 2 cm or larger [[21]]. Therefore, the nodule size was classified into 3 categories in our study. TIRADS had best value in nodules ≤10 mm, while CEUS performed best for differentiating lesions >10 mm, especially in lesions ≥20 mm yield higher specificity.

Under US, TIRADS has been suggested as a helpful criterion in the diagnosis of malignant thyroid nodules with the sensitivity, specificity, PPV, NPV were 88%, 80%, 79%, and 88%, respectively, in a study with 1097 nodules [[22]]. The malignancy risk of thyroid nodules could be stratified into the simplified five categories of TIRADS according to 10 US patterns [[23]]. The malignancy rates of TIRADS score 3, score 4 were 14.6, and 48.9 %, respectively, equal to the recommended range (3∼15%, 15∼50%, respectively). Meanwhile, with the criteria of TIRADS for nodule size was 10–20 mm in our study, the sensitivity and PPV were 84.7% and 78.1%, respectively, which were comparable to those of Zhang's study [[19]]. With the criteria of TIRADS for nodules ≤10 mm, the sensitivity and PPV were 81.8% and 72.0%, respectively, and with the criteria of TIRADS for nodules ≥20 mm, those were 81.5% and 70.5%, respectively, all pretty lower than the results of 10–20 mm. A recent prospective validation of TIRADS on 3980 thyroid nodules found that the sensitivity and specificity was 0.85 and 0.89, which was equal to our results [[24]]. Our data suggested that the overall diagnostic performance of TIRADS was a good US method. Therefore, although TIRADS was based on nodules measured >10 mm, it was also applicable to nodules less than 10 mm. However, some nodules in our study didn't meet the criteria of TIRADS classification, including some patterns of partial cyst [[25]], such as the proportion of solid components of partially cystic thyroid nodules cannot predict a malignancy [Fig. 1 A, E]. Thus, FNA for these nodules should be applied.

Graph: Fig.1 A 44-year-old woman with two thyroid nodules: a 2.8×2.0 cm solid nodule in the upper right lobe and a 0.7×0.7 cm solid nodule in the lower right lobe. (A) Conventional US found four malignant indicators (solid, markedly hypoechoic, microcalcifications and aspect ratio = 1) in smaller one, and it was classified to 5 TIRADS score. (B) Conventional US found no suspicious feature in larger one, and it was classified to 3 TIRADS score. (C) A CEUS indicated hyper-enhancement in larger one and hypo-enhancement in smaller one. (D) The pathological image of the larger nodule, which was of a follicular thyroid carcinoma. (E) The pathological image of the smaller nodule, which was of a nodular goitre.

CEUS has been reported to improve the identification of malignant thyroid nodules [[26]]. Previous studies have also demonstrated that, under CEUS, rim-like enhancement correlated highly with a benign diagnosis (sensitivity 83.0%, specificity 94.1%, and accuracy 88.5%) and hypo-enhancement correlated highly with a malignant diagnosis (sensitivity 88.2%, specificity 92.5%, and accuracy 90.4%) [[28]]. In our study, the malignancy risks were 66.7% for the hypo-enhancement mode, 14.8% for the iso-enhancement mode, 5.1% for the hyper-enhancement mode, and 6.5% for the rim-like enhancement mode. Our study demonstrated that most malignant nodules had hypo-enhancement and benign thyroid nodules had hyper-enhancement and rim-like enhancement using CEUS. Most malignant nodules have fibrosis, calcification, focal necrosis, and aberrant blood vessels, which may lead to hypo-enhancement on CEUS [[29]]. However, our study also found that some nodules were unable to be reliable to the conventional CEUS enhancement patterns, including papillary thyroid microcarcinoma, which enhancement patterns varied between benign and malignant lesions [Fig. 1 C]. Xu et al. [[30]] reported that combined CEUS with shear wave elastography would be help to improve diagnostic accuracy of papillary thyroid microcarcinoma.

At present, there is no conclusive conclusion about the enhancement patterns of thyroid nodules ≤10 mm. Numerous previous studies found that the enhancement patterns of thyroid nodules ≤10 mm are various and have no consistent features. In our study, CEUS in nodules ≤10 mm had a lowest AUC and specificity among the three groups, in line with the conclusions of Bartolotta et al., who reported that nodules measuring <1 cm showed mainly absent vascularization [[31]]. Jebreel et al. [[32]] reported that differences in the microvessel density of benign nodules ≤10 mm and malignant nodules may not be evident. Furthermore, malignant nodules ≤10 mm and benign lesions may exhibit different in vascular diameter and shape, vascular branches, arteriovenous fistulae, new tumor vessels, or edema and fbrosis of the lesion stroma [[33]]. In group A, the differences in diagnostic value between TIRADS and CEUS were significant (AUC: 0.804 vs 0.733, P = 0.028, sensitivity: 81.8% vs 72.7%, P = 0.013, specificity: 88.9% vs 79.4%, P = 0.011). According to the TIRADS based on each suspicious US feature, we found that the AUC, sensitivity, specificity of nodules <10 mm in size were 0.804, 81.8% and 88.9%, respectively, which higher than those of CEUS. A study by Ma found that ultrasound features are important predictors of nodules <10 mm, such as taller-than-wide shape, marked hypoechogenicity and poorly defined margin of the nodule [[34]]. Our data demonstrated that conventional US is still the most important US tool in evaluating nodules <10 mm.

Another novel finding in this study was that nodule diameter ≥20 mm also influenced the diagnostic performance of the TIRADS and CEUS, and the specificity of CEUS in nodules ≥20 mm was significantly higher compared with TIRADS classification (90.8% vs 82.9%, P = 0.023). Conflicting results have been reported regarding the association of nodule size >2 cm and a higher prevalence of malignancy overall. Jennifer et al. [[35]] reported that large (≥2 cm) nodules have a higher pretest probability of malignancy, while a retrospective analysis of prospective database showed that thyroid nodule size ≥4 cm is not associated with a higher malignancy risk [[36]]. In our study, the malignancy rate of thyroid nodule size ≥20 mm was 33.3% (38 of 114), which was equal to Hong et al. [[4]] reported 33.5%. In those malignant nodules, there were 6 follicular thyroid carcinoma, which always showed ovoid in shape, minor calcification and smooth in boundary under US, and were classified as low- (TIRADS 3) and intermediate- (TIRADS 4) suspicion US features. However, except 1 case [Fig. 1 B∼D], the rest of follicular thyroid carcinoma (83.3%, 5 of 6) have destruction of blood vessels, losing its function and tumor thrombus forming, which may lead to hypo-enhancement on CEUS. Therefore, combined CEUS with TI-RADS would be help to improve diagnostic accuracy of follicular thyroid carcinoma.

The limitations of this study are as follows: the patients in this study were selected for FNA or surgery; as such, a selection bias was inevitable. Secondly, in subsequent studies, the sample size should be enlarged to allow for more accurate calculation of sensitivity, specificity and predictive values. Finally, this was a retrospective single-center study, prospective larger multicenter studies with a nonsurgical population are mandatory in the future.

In summary, both TIRADS and CEUS patterns provide effective assessment criterion for the discrimination of thyroid nodule. Nodule size influences the diagnostic accuracy of the two methods. TIRADS had best value in nodules ≤10 mm, while CEUS performed best for differentiating lesions >10 mm, especially in lesions ≥20 mm.

None.

We did not receive any research funds for this study. The authors thank all of the patients who participated in this study.