Analysis and prediction of contralateral central lymph node metastasis risk in unilateral papillary thyroid carcinoma with ipsilateral lateral cervical lymph node: a retrospective clinical study

Abstract

Background: Papillary thyroid carcinoma (PTC) often metastasizes to lymph nodes, increasing recurrence risk and reducing survival. This study identifies predictors for contralateral central lymph node metastasis (Cont-CLNM) in unilateral PTC patients with ipsilateral lateral cervical lymph node metastasis (Ipsi-LLNM).

Methods: We retrospectively analyzed data, preoperative ultrasound features, and thyroglobulin (Tg) levels in unilateral PTC patients with Ipsi-LLNM treated at the Thyroid Surgery Department of Nanjing Drum Tower Hospital from August 2017 to August 2024. Least absolute shrinkage and selection operator (LASSO) regression was used for variable selection, with independent t-tests and Chi-squared tests assessing differences. Logistic regression analyses identified risk factors for Cont-CLNM, and a nomogram was validated using 1,000 bootstrap resamples. Decision curve analysis (DCA) evaluated clinical impact.

Results: Of 105 PTC patients, 56 (53.3%) had Cont-CLNM. LASSO regression identified three predictors: male sex, lymph node metastasis posterior to the recurrent laryngeal nerve (LN-prRLN), and elevated Tg levels. Multivariate regression confirmed these variables' association with Cont-CLNM. Internal validation yielded an area under the curve of 0.771 [95% confidence interval (CI): 0.684-0.857]. A nomogram was developed and validated through DCA.

Conclusions: Our findings indicate that combining male gender, LN-prRLN, and Tg levels effectively predicts Cont-CLNM, providing a basis for risk assessment in unilateral PTC.

Keywords: Papillary thyroid carcinoma (PTC); contralateral central lymph node metastasis (Cont-CLNM); ipsilateral lateral cervical lymph node metastasis (Ipsi-LLNM); predictive model; risk factors.

Figure 1

LASSO regression screening for factors affecting the occurrence of Cont-CLNM in Ipsi-LLMN-PTC. (A) LASSO coefficient path for study variables; (B) selection of optimal λ via 10-fold cross-validation. (B) Three predictor variables with non-zero coefficients were identified as the key factors influencing contralateral central lymph node metastasis in LLNM-PTC patients. These variables resulted in the smallest error and the best model fit. The predictor factors identified from the left figure, corresponding to the intersection with the lambda =3 line, include gender, lymph node metastasis of the recurrent laryngeal nerve, and thyroglobulin levels. Cont-CLNM, contralateral central lymph node metastasis; Ipsi-LLMN-PTC, papillary thyroid carcinoma patients with ipsilateral lateral cervical lymph node metastasis; LASSO, least absolute shrinkage and selection operator; LLNM-PTC, papillary thyroid carcinoma patients with lateral cervical lymph node metastasis; LN-prRLN, lymph node metastasis posterior to the recurrent laryngeal nerve.

Figure 2

Nomogram of clinical prediction model for factors affecting the occurrence of Cont-CLNM in Ipsi-LLMN-PTC. (A) A nomogram developed using three predictive variables selected through LASSO. (B) Expanding on this model by incorporating tumor size as an additional predictive factor. The length of each variable’s line segment reflects its relative contribution to the clinical outcome. The total score, calculated as the sum of individual variable scores, corresponds to a probability scale that estimates the risk of Cont-CLNs metastasis in LLNM-PTC patients. Cont-CLNs, contralateral central lymph nodes; Ipsi-LLMN-PTC, papillary thyroid carcinoma patients with ipsilateral lateral cervical lymph node metastasis; LASSO, least absolute shrinkage and selection operator; LLNM-PTC, papillary thyroid carcinoma patients with lateral cervical lymph node metastasis; LN-prRLN, lymph node metastasis posterior to the recurrent laryngeal nerve; Tg, thyroglobulin.

Figure 3

Decision curve analysis of clinical prediction model for Cont-CLNM in Ipsi-LLMN-PTC. The red line (LASSO) represents the model established using the three predictor variables selected through LASSO screening. The blue line (More) represents the model built upon these three variables with the addition of tumor size as a predictor. The LASSO model consistently demonstrated significantly higher net benefits across all threshold ranges compared to the More model and no or full intervention. All, full intervention; Cont-CLNM, contralateral central lymph node metastasis; Ipsi-LLMN-PTC, papillary thyroid carcinoma patients with ipsilateral lateral cervical lymph node metastasis; LASSO, least absolute shrinkage and selection operator; More, the More model; None, no intervention.

Figure 4

Clinical prediction model for the risk of Cont-CLNM in Ipsi-LLMN-PTC: results of bootstrap internal sampling validation. (A) ROC curves from 1,000 internal bootstrap resamples. The model’s calibration AUC is 0.771, with a 95% confidence interval of 0.684–0.857, indicating good fit. (B) Bootstrap calibration curve. AUC, area under the curve; Cont-CLNM, contralateral central lymph node metastasis; Ipsi-LLMN-PTC, papillary thyroid carcinoma patients with ipsilateral lateral cervical lymph node metastasis; ROC, receiver operating characteristic.