MRI-based model to predict preoperative extrathyroidal extension in papillary thyroid carcinoma

This study aimed to develop and validate a predictive model for preoperative extrathyroidal extension (ETE) in papillary thyroid carcinoma (PTC) using MRI features.

We retrospectively analyzed 140 confirmed PTC cases, divided into training (n = 84) and validation (n = 56) groups. MRI features such as T2-weighted imaging, multiphase contrast-enhanced MRI, and diffusion-weighted imaging were evaluated along with clinical data. Univariate and multivariate logistic regression identified independent predictors of ETE and developed a predictive nomogram. We evaluated the nomogram's discrimination, calibration, and clinical utility, and performed subgroup analyses to explore the relationships between risk factors and baseline data. Predictive performance was assessed using ROC curves and DeLong tests.

Age, protrusion value, and apparent diffusion coefficient_Brightest_rate (ADC_Best_rate) were independent predictors of ETE. The nomogram effectively differentiated ETE from no-ETE, showing strong discrimination, clinical utility, and calibration in both the training (AUC = 0.826, Hosmer-Lemeshow p = 0.882) and validation cohorts (AUC = 0.805, Hosmer-Lemeshow p = 0.585). The model performed consistently across different MRI systems (1.5 T and 3.0 T) and gender subgroups. Notably, ADC_Best_rate (AUC = 0.742) outperformed ADC_mean_rate and ADC_minimum_rate. A significant interaction between ADC_Best_rate and gender (p = 0.02) showed that ADC_Best_rate predicted ETE in PTC more accurately in males (AUC = 0.897) compared to females (AUC = 0.644).

Our nomogram model, incorporating age, protrusion value, and ADC_Best_rate, effectively predicted preoperative ETE in PTC patients, aiding surgeons in optimizing therapeutic decision-making. ADC_Best_rate may be a promising potential indicator in MRI functional imaging.

Question This study addresses the challenge of accurately predicting extrathyroidal extension (ETE) in papillary thyroid carcinoma (PTC) to improve surgical decision-making. Findings A predictive nomogram incorporating age, protrusion value, and ADC_Best_rate effectively differentiates ETE from no-ETE, showing strong performance in both training and validation cohorts. Clinical relevance This nomogram aids surgeons in identifying patients at risk for ETE, enhancing therapeutic decision-making and potentially improving patient outcomes in PTC management.