In patients with nodules that are highly suspected of malignancy in one thyroid lobe preoperatively, accurate assessment of the presence of malignant nodules in the contralateral lobe by ultrasound (US) is critical, which will directly influence the surgical scope. However, US has limitations in the risk stratification of malignant nodules in the contralateral lobe[13, 14]. Therefore, how to improve the ability of contralateral nodule malignancy stratification risk in papillary thyroid carcinoma (PTC) patients should be concerned. In our study, we found three parameters that can modify the risk of malignancy of contralateral nodules, including body mass index (BMI) ≥ 25 kg/m2, multifocality present in suspected-lobe and isthmus nodule with Thyroid Imaging Report and Data System of the American College of Radiology (ACR TI-RADS) classification > 3 (P < 0.05).
Obesity is recognized as a risk factor for the development of thyroid cancer[20]. O'Neill et al. reported that the occurrence of multifocal PTC was related to the increase of BMI[21], and with the increase of BMI, the incidence of multifocality increased. Bilateral lobe incidence of multifocal cancer in our study is also consistent with this trend. The mechanism of the internal relationship between increased BMI and multifocal PTC is still unclear presently, and it was reported insulin resistance and hyperglycemia caused by obesity can trigger tumor transduction pathway and promote the proliferation of tumor cells[22].
Some studies have shown that multifocality is a risk factor for contralateral thyroid cancer[23, 24]. Consistent with these findings, our study found a higher proportion of contralateral cancers when the suspected-lobe was multifocal. It is believed that multifocality is the result of monoclonal metastasis within the gland of thyroid tumors[25]. The human thyroid gland has a rich lymphatic vessels system, which run through the whole thyroid gland and connect with each other to form a network[26], and the interconnected network-like structure may be the anatomical basis of multifocal development. Therefore, we speculate that when multiple cancer occurs in the one lobe, intraglandular dissemination in the contralateral lobe may occur at the same time, which may explain the internal relationship between the one-side multiple foci and contralateral cancers.
The thyroid isthmus has a special anatomical position, which locates in front of the trachea and connects bilateral lobes of thyroid gland. Although the incidence of PTC in the isthmus is low, some studies have shown that isthmus PTC is more aggressive than lateral lobe PTC and is associated with the occurrence of multifocal lesions[27–29]. This phenomenon may be explained by the anatomical characteristics of the isthmus. Referenced with immunohistochemical-labeled pathological specimens from patients with PTC, we plotted a hypothetical picture of thyroid isthmus tumor spreading through the intraglandular lymphatic network (Fig. 5). Because the isthmus located in the center of the thyroid, the cancer cell in the isthmus can spread to lateral lobes[30], which may be caused by the spread of intraglandular lymphatic network. Therefore, we believe that the isthmus lesion should be concerned. Our study showed that ACR TI-RADS classification > 3 may be one of the parameters associated with contralateral PTC. However, it should be noted that due to the limitation of growth space, the US appearance of tumor located in the isthmus may not be typical[31], so US evaluation of the isthmus lesion should be more careful.
In our study, the ACR TI-RADS classification of contralateral lobe nodules needed to be upgraded when the patient had BMI ≥ 25kg/m2, the presence of multifocality in suspected-lobe and nodule in the isthmus with ACR TI-RADS classification > 3. By adding these parameters, we can obtain more information about the contralateral lobe nodules, rather than focusing only on the limited ultrasonographic parameters of TI-RADS. After modifying the malignant risk stratification of contralateral nodules, the AUC for diagnosing contralateral malignant nodules increased from 0.796 (95%CI: 0.743–0.842) to 0.830 (95%CI: 0.780–0.872). We also need to pay attention to occult cancer in clinical practice. It has been reported that the incidence of occult PTC in the contralateral lobe is 12% − 40%[32–34]. After adjustment with our method, the detection ability of occult carcinoma in the thyroid contralateral lobe was significantly improved (the missed diagnosis rate decreased from 21.1–4.5%), and this result was validated in the validation set (the missed diagnosis rate decreased from 11.4–2.9%).
Tumor size can be easily measured by preoperative US, which is a common characteristic for US evaluation of thyroid nodules. However, our results showed that the presence of contralateral PTC was not related to the size of the suspected lesion, which is consistent with the findings of Pitt et al.[35]. Therefore, the relationship between tumor size and contralateral cancer needs to be further investigated.
The main ultrasonographic features of malignant thyroid nodules are taller-than-wide shape, irregular margins, solid, marked hypoechogenicity and microcalcification[36, 37]. However, ultrasonographic features of benign and malignant nodules often have some overlaps, which is easy to cause inaccurate US TI-RADS classification. By comparing the differences in ultrasonographic features between the underestimated and un-underestimated contralateral PTC, we found that the presence of cystic components in the nodules and the surrounding halo were more likely to be underestimated. Although a presence of cystic components is often seen in benign nodules, there is still a risk of malignancy in mixed cystic nodules. It has been reported that the malignant risk of nodules with cystic part ≤ 10% is 8.8%[38]. In this study, the PTC with cystic component in the contralateral lobe was underestimated. Therefore, it should not be degraded easily when a cystic component is present in a suspected malignant nodule, especially when PTC has been confirmed in one lobe. At present, there are different opinions on the diagnostic value of halo, which is not included in the evaluation system of ACR TI-RADS. Some studies have suggested that thin halo can reduce the risk of tumor malignancy, while malignant nodules often have an absent halo or have thick halo[39]. However, there is no unified standard for distinguishing thick halo from thin halo, thus it is subjective to use this marker to distinguish benign nodules from cancers. Siebert et al. reported a special halo performance, which was characterized as a non-uniform hypoechoic rim circumferentially around the nodule central hyperechoic area[40], and similar findings were observed in our cases. Siebert et al. believed that this feature was significantly related to PTC with a specificity of 91.3%, the central hyperechoic area was scarring and conglomerate fibrosis, and the surrounding area was papillary carcinoma in the pathological specimens. Therefore, we believe that standardized US definition of the malignant halo of nodules may be an effective method to avoid underestimation, which needs further study. In addition, it was interesting to observe that two patients had significantly different ultrasonographic features in the unilateral lobe and the contralateral lobe PTC in the underestimated group (Fig. 3B), which may be explained by the different origins of the multifocal clones. The occurrence of multifocal foci may be based on coexistence of two patterns of clonal homologous and clonal independent origin[41, 42], so tumors from the same patient may present different ultrasonographic findings. Regarding occult cancer, our study showed that when the pathological size of tumor was less than 0.5 cm, it was more likely to missed by US. Although US can detect small lesions about 0.2cm in thyroid, these small lesions may co-exist with nodular goiter or Hashimoto's thyroiditis, which may be the reason why small lesions of 0.2-0.5cm are still difficult to be detected by US.
This study has some limitations. Firstly, this study was a single-center retrospective study, which may have selection bias. It is necessary to conduct a prospective multi-center study to verify our results. Secondly, due to the limitation of inclusion criteria, the sample size of this study was limited, and it is still necessary to increase the sample size for further research.
In conclusion, the diagnostic efficiency of contralateral lobe malignant nodules was improved by modified ACR TI-RADS classification, which may assist surgeons to select resection scope and avoid residual lesions. Furthermore, nodules with cystic components or halo in the contralateral lobe may tend to be underestimated, which should attract more attention in preoperative US practice.