Risk factors and prediction model of level II lymph node metastasis in papillary thyroid carcinoma.

Introduction: Surgical management of lateral lymph nodes in papillary thyroid carcinoma, especially at level II, remains controversial. This study aimed to investigate the risk factors for level II lymph node metastasis in patients with papillary thyroid carcinoma and establish a prediction model to estimate the metastatic risk. Materials and methods: A total of 768 patients with papillary thyroid carcinoma underwent thyroidectomy and central plus lateral lymph node dissection, including levels VI, II, III, and IV, at the First Affiliated Hospital of Chongqing Medical University from January 2016 to December 2018. Data on the clinicopathological characteristics were collected and analyzed. Univariate and multivariate analyses were performed to identify risk factors for level II lymph node metastasis. Subsequently, a predictive model was established based on the results of the multivariate analyses. Results: The level II lymph node metastatic rate was 34.11% with the following features: largest tumor diameter >20 mm (Odds ratio=1.629, P=0.026), located in the upper pole (Odds ratio=4.970, P<0.001), clinical lymph node-positive (clinical central lymph node-positive: Odds ratio=1.797; clinical lateral lymph node-positive: Odds ratio=1.805, P=0.008), vascular invasion (Odds ratio=6.759, P=0.012), and rate of central lymph node metastasis (Odds ratio=2.498, P<0.001). Level III lymph node metastasis (Odds ratio=2.749, P<0.001) and level IV lymph node metastasis (Odds ratio=1.732, P=0.007) were independent of level II lymph node metastasis predictors. The prediction model's areas under the receiver operating characteristic curve were 0.815 and 0.804, based on bootstrapping validation. Level II lymph node metastasis was associated with the tumor-free survival rate of patients with papillary thyroid carcinoma (P<0.001). Conclusions: Largest tumor diameter >20 mm, located in the upper pole, clinical lymph node-positive, vascular invasion, rate of central lymph node metastasis, and levels III and IV lymph node metastases were independent level II lymph node metastasis predictors. We developed a prediction model for level II lymph node metastasis. Overall, level II lymph node metastasis dissection should be individualized according to clinicopathological data both preoperatively and intraoperatively.