Nomogram model based on preoperative serum thyroglobulin and clinical characteristics of papillary thyroid carcinoma to predict cervical lymph node metastasis

Abstract

Objective: Preoperative evaluation of cervical lymph node metastasis (LNM) in papillary thyroid carcinoma (PTC) has been one of the serious clinical challenges. The present study aims at understanding the relationship between preoperative serum thyroglobulin (PS-Tg) and LNM and intends to establish nomogram models to predict cervical LNM.

Methods: The data of 1,324 PTC patients were retrospectively collected and randomly divided into training cohort (n = 993) and validation cohort (n = 331). Univariate and multivariate logistic regression analyses were performed to determine the risk factors of central lymph node metastasis (CLNM) and lateral lymph node metastasis (LLNM). The nomogram models were constructed and further evaluated by 1,000 resampling bootstrap analyses. The receiver operating characteristic curve (ROC curve), calibration curve, and decision curve analysis (DCA) of the nomogram models were carried out for the training, validation, and external validation cohorts.

Results: Analyses revealed that age, male, maximum tumor size >1 cm, PS-Tg ≥31.650 ng/ml, extrathyroidal extension (ETE), and multifocality were the significant risk factors for CLNM in PTC patients. Similarly, such factors as maximum tumor size >1 cm, PS-Tg ≥30.175 ng/ml, CLNM positive, ETE, and multifocality were significantly related to LLNM. Two nomogram models predicting the risk of CLNM and LLNM were established with a favorable C-index of 0.801 and 0.911, respectively. Both nomogram models demonstrated good calibration and clinical benefits in the training and validation cohorts.

Conclusion: PS-Tg level is an independent risk factor for both CLNM and LLNM. The nomogram based on PS-Tg and other clinical characteristics are effective for predicting cervical LNM in PTC patients.

Keywords: lymph node metastasis; nomogram; papillary thyroid carcinoma; preoperative serum thyroglobulin; surgery.

Figure 1

Enrollment flowchart of participants for model development and validation. PTC, papillary thyroid carcinoma; PS-Tg, preoperative serum thyroglobulin.

Figure 2

The ROC curve and optimal cutoff value of PS-Tg for CLNM diagnosis (A) and LLNM diagnosis (B). PS-Tg, preoperative serum thyroglobulin; CLNM, central lymph node metastasis; LLNM, lateral lymph node metastasis.

Figure 3

The nomograms indicate the risk of CLNM (A) and LLNM (B) based on clinical factors. CLNM, central lymph node metastasis; LLNM, lateral lymph node metastasis; TSH, thyroid stimulating hormone; PS-Tg, preoperative serum thyroglobulin; ETE, extrathyroidal extension.

Figure 4

ROC curve showing prediction effective of the nomogram model for CLNM in the training group (A), the internal validation group (B), and the external validation group (C), as well as nomogram model for LLNM in the training group (D), the internal validation group (E), and the external validation group (F). CLNM, central lymph node metastasis; LLNM, lateral lymph node metastasis.

Figure 5

Calibration curve of the prediction nomogram for CLNM in the training group (A), the internal validation group (B), and the external validation group (C), as well as nomogram model for LLNM in the training group (D), the internal validation group (E), and the external validation group (F). The x-axis represents the predicted probability of CLNM or LLNM, while the y-axis stands for the actual diagnosed probability of CLNM or LLNM. The diagonal dashed line represents an ideal prediction model. The solid line shows the performance of the nomogram models, of which a closer fit to the diagonal dashed line indicates better prediction ability. CLNM, central lymph node metastasis; LLNM, lateral lymph node metastasis.

Figure 6

Decision curve of the prediction nomogram for CLNM in the training group (A), the internal validation group (B), and the external validation group (C), as well as nomogram model for LLNM in the training group (D), the internal validation group (E), and the external validation group (F). The red line represents the nomogram model. The gray line represents the assumption that all patients are CLNM/LLNM positive, while the horizontal black line represents that all patients are CLNM/LLNM negative. CLNM, central lymph node metastasis; LLNM, lateral lymph node metastasis.