Development and validation of a collagen signature to predict central lymph node metastasis in papillary thyroid cancer

Background Current clinicopathological risk factors lack the precision necessary for accurate prediction of central lymph node metastasis (CLNM) in patients with papillary thyroid cancer (PTC). Structural remodeling of the tumor microenvironment (TME), particularly collagen organization, may play a pivotal role in metastatic dissemination. Objective The objective of this study was to develop a collagen signature within the TME to predict CLNM in PTC and validate that the new model incorporating it into the assessment alongside clinicopathological risk factors would enhance the predictive accuracy. Methods In this retrospective study, we included 350 patients with classic PTC, all of whom underwent thyroidectomy with prophylactic central lymph node dissection. The cases were randomly assigned to a training cohort and a testing cohort with a 6:4 ratio. A total of 142 collagen features in the TME were extracted from second harmonic generation images of tumor specimens. We constructed a collagen signature using a least absolute shrinkage and selection operator (LASSO) regression model. Multivariate logistic regression was used to integrate the signature with clinicopathological variables and construct a nomogram. Results The predictive ability of collagen signature was also validated by AUC of 0.821 in training cohort and AUC of 0.793 in testing cohort. The collagen signature remained an independent predictor after adjustment for tumor size, capsular invasion, and tumor location in the multivariate analysis. Furthermore, the integrated model showed superior predictive performance compared to the clinicopathological model alone (0.842 vs. 0.679, p 0.001). Decision curve analysis confirmed higher net clinical benefit across a wide range of thresholds. Conclusions The collagen signature within the TME represents a promising new biomarker that can effectively predict CLNM in PTC patients, potentially improving clinical decision-making and patient management.