Recent epidemiological studies have suggested a real increase in the incidence of LATC patients [2, 19]. The stark contrast between poor prognosis and increased incidence underscores the critical need for heightened clinical attention to patients with locally advanced disease. However, current literature on locally advanced thyroid cancer is predominantly based on small sample cohorts, which may introduce potential selection bias and confounding factors [6, 7, 10, 12, 16, 20, 21]. This study is the first to utilize the SEER database, a nationwide large sample cohort, to provide comprehensive data on the factors predicting locally advanced invasion and influencing invasion patterns in LATC patients. Furthermore, this study explored the significant different survival outcomes across various invasion sites and established a novel site-based risk stratification based on 8th AJCC T staging for locally advanced PTC patients, which demonstrated better discrimination than 8th AJCC T staging.

Previous researches have reported that LATC patients with R0 resection (no tumor residue) or R1 resection (microscopic tumor residue) have better prognoses than those with R2 resection (macroscopic tumor residue) [10, 12]. One of the challenges in surgical intervention is distinguishing LATC patients and their invasion sites preoperatively to avoid R2 resection due to inefficient preoperative preparation. Ultrasound is the most common examination for screening suspicious LATC patients and consequent computerized tomography or magnetic resonance imaging are employed in these patients to identify specific invasion sites. However, considering the sensitivity and specificity of ultrasound, understanding the factors predicting locally advanced invasion and influencing invasion patterns may help decide further imaging examinations, find specific invasion sites and improve surgery performance. Xu et al. suggested that the distribution of age, sex, tumor size, N stage and M stage significantly changed from no extrathyroidal extension to extrathyroidal extension [22]. Similarly, this study further found that patients aged≥65 years (HR = 1.91, p < 0.001), those with tumor size ≥40 mm (HR = 2.48, p < 0.001), those with lymph node metastasis (HR = 1.32, p < 0.001), those with distant metastasis (HR = 1.52, p < 0.001) and those with MTC (HR = 1.31, p < 0.001) or ATC (HR = 1.56, p < 0.001) were more likely to have locally advanced invasion progressing from strap muscle invasion. These findings indicate that more attention and further imaging examinations should be paid to potential LATC patients with these characteristics.

Wang et al. investigated the invasion patterns in locally advanced differentiated thyroid carcinoma patients, and found that recurrent laryngeal nerve and trachea invasion were most common [12]. Our study further explored the influence of thyroid cancer subtypes, tumor size, age and sex on invasion patterns. Consistent with Wang’s findings, our results suggest that for the management of PTC patients, potential parathyroid or nerve invasion and trachea invasion should be emphasized, while potential blood vessel invasion should be more monitored in FTC/OTC patients and ATC patients. Additionally, elderly patients, those with large tumors, and male patients demonstrated higher rates of trachea invasion, rather than isolated parathyroid or nerve invasion. Integration of these parameters in suspicious LATC patients can inform the focus of imaging examinations and guide the selection of surgical procedures and comprehensive treatment planning. This may mitigate risks of intraoperative cessation, tumor residual (due to R1/R2 resection), or increased complication rates resulting from missed diagnoses of locally advanced invasion in clinical practice [14, 15, 23].

According to the 8th AJCC T staging system, invasion of subcutaneous soft tissues, larynx, trachea, esophagus, or recurrent laryngeal nerve invasion is classified as T4a, while tumor encasement of the carotid artery or mediastinal vessels is defined as T4b. Some clinicians consider this risk stratification inadequate, as they have observed significantly different survival outcomes associated with various invasion sites and it does not take multi-invasion into account. Abraham et al. found that patients with multi-invasion had a fivefold increased risk of recurrence compared to those with single-site invasion [17]. Moreover, previous studies attempting to demonstrate the prognosis of specific invasion sites, but they reported significantly conflicting results, likely due to selection bias and confounding effects [6, 16, 18]. For instance, Sessa et al. reported a 5-year disease-free rate of approximately 65% for patients with trachea invasion, while Song et al. found it to be 80% [6, 16]. Our study is the first to utilize a large sample cohort to reveal prognostic variations by different invasion sites in locally advanced PTC patients. Our findings illustrated that almost every invasion site is associated with significantly different survival outcomes, suggesting the necessity for a new, detailed site-based risk stratification system. After adjusting for confounding factors by multivariable Cox regression, all LATC patients could be stratified into four risk layers based on CSS rates: (1) Extremely low risk: parathyroid or nerve invasion (15-year CSS rate: 88.8%); (2) Low risk: bone or skeletal muscle invasion and esophagus or larynx invasion (15-year CSS rate: 79.3%); (3) Medium risk: trachea invasion (15-year CSS rate: 68.9%); (4) High risk: blood vessel invasion and multi-invasion (15-year CSS rate: 57.1%). Survival curves comparing the 8th AJCC T staging and this new risk stratification showed that new site-based risk stratification offers better discrimination. This novel stratification enables tailored treatment strategies based on preoperatively identified invasion sites, with more aggressive approaches considered for sites associated with poorer prognosis. For instance, the implementation of neoadjuvant targeted or immunological therapy may be more strongly considered in patients with trachea invasion, blood vessel invasion, or multi-invasion to potentially improve prognostic outcomes in these patients.

Our study has several limitations. Firstly, since SEER database only registers initial LATC patients, the prognosis of patients with recurrent LATC, who may be more common in clinical practice, could not be assessed. Further studies containing both initial and recurrent LATC patients should be conducted in the real world to validate our findings. Secondly, our results revealed that LATC showed a significantly increased rate of distant metastasis, especially in patients with locally advanced FTC/OTC, MTC, and ATC. These patients may have unique prognostic profiles, but they were not the focus of this study. Further relevant studies are needed to address this subgroup. Thirdly, due to the limitation of SEER records (‘CS extension’), the calculation of parathyroid or nerve invasion rates in our results only include patients with isolated parathyroid or nerve invasion (refer to the methods section), which may underestimate the true proportion of parathyroid or nerve invasion (refer to previous studies [6, 7, 12]). Finally, the lack of information about invasion depth in the SEER database made it difficult to further stratify prognosis among specific invasion sites, particularly in patients with trachea invasion. Future studies containing detailed invasion depth and different surgical procedures in single invasion site should be conducted.