# CD31-associated vascular phenotyping using Doppler ultrasound and dual-energy CT for recurrence risk stratification in papillary thyroid cancer

## TL;DR

This study uses Doppler ultrasound and dual-energy CT to assess blood vessel patterns linked to CD31 in thyroid cancer patients, helping predict recurrence risk.

## Contribution

A novel multimodal imaging framework for noninvasive CD31-associated vascular phenotyping in papillary thyroid cancer.

## Key findings

- High CD31 expression correlates with advanced tumor stage, metastasis, and recurrence.
- The vascular phenotype model achieved an AUC of 0.815 for predicting CD31 expression.
- The imaging-derived vascular phenotype significantly predicted recurrence-free survival across datasets.

## Abstract

Angiogenesis plays a pivotal role in PTC aggressiveness and recurrence. CD31, an endothelial marker, provides a histological correlate of tumor vascularity. This study aimed to establish a multimodal imaging framework using Doppler ultrasound (US) and dual-energy CT (DECT) to capture CD31-associated vascular phenotypes for recurrence risk stratification.

A total of 414 PTC patients (training set, 151; test set 1, 137; test set 2, 126) were included. Clinical and radiographic features, including demographics, tumor morphology, Doppler US grade, US grayscale values, and DECT parameters, were analyzed. CD31 expression was assessed by immunohistochemistry in the training set and served as a reference for vascular phenotyping. Logistic regression was applied to identify imaging-derived vascular signatures linked to CD31 expression. Recurrence-free survival (RFS) was analyzed across all datasets to evaluate prognostic value.

High CD31 expression correlated with advanced tumor stage (p = 0.003), lymph node metastasis (p = 0.001), extrathyroidal extension (p < 0.001), and recurrence (p < 0.001). The vascular phenotype model integrating Doppler US grade, unenhanced electron density (ED), arterial-phase iodine concentration (IC), and venous-phase IC/ED ratio achieved an AUC of 0.815 (95% CI: 0.744–0.874) for discriminating CD31 expression in the training set. This imaging-derived phenotype was significantly associated with RFS in the training set (p = 0.017), test set 1 (p = 0.006), and test set 2 (p = 0.002), effectively identifying patients at higher risk of recurrence.

Preoperative vascular phenotyping with Doppler US and DECT enables noninvasive assessment of CD31-associated angiogenic activity and enhances recurrence risk stratification in PTC.

The online version contains supplementary material available at 10.1186/s40644-025-00975-w.