Diagnostic accuracy of combined TCF-21 protein expression and computed tomography morphological features for...

Preoperative identification of microvascular invasion (MVI) in clinically early-stage non-small cell lung cancer (NSCLC) remains difficult because microscopic vascular spread cannot be reliably inferred from tumor size or nonspecific computed tomography (CT) features alone. This diagnostic accuracy study aimed to evaluate whether biopsy-derived transcription factor 21 (TCF-21) protein expression combined with a rigorously defined CT vessel convergence sign could predict pathological MVI.

A retrospective diagnostic accuracy cohort of 50 patients with pathologically confirmed stage I NSCLC who underwent curative-intent surgical resection at the Affiliated Hospital of Hebei University of Engineering between March 2023 and August 2023 was enrolled. Patients were categorized into MVI-positive (n=15) and MVI-negative (n=35) groups according to the postoperative pathological reference standard. Preoperative chest CT scans and TCF-21 protein expression levels quantified by Western blot from preoperative biopsy specimens were evaluated. CT morphological features included tumor size, spiculation, lobulation, and vessel convergence sign, which was strictly defined to exclude volume-loss atelectasis or mechanical vascular displacement. Logistic regression and receiver operating characteristic (ROC) curve analyses were used to develop and evaluate the exploratory prediction model.

The cohort predominantly consisted of men (62.0%), and the mean age was 55.9 years. TCF-21 protein expression was significantly lower in the MVI-positive group than in the MVI-negative group (0.42±0.11 vs. 0.85±0.15, P<0.001). A distinct CT vessel convergence sign was more frequent in MVI-positive tumors than in MVI-negative tumors (11/15, 73.33% vs. 8/35, 22.86%; P<0.001), despite no significant between-group difference in tumor diameter. Spearman correlation indicated a significant negative association between TCF-21 expression and MVI status (r=-0.582, P<0.001). The combined model demonstrated the highest apparent diagnostic performance, with an area under the ROC curve of 0.892 [95% confidence interval (CI): 0.795-0.989], sensitivity of 86.67% (13/15), and specificity of 82.86% (29/35).

Decreased TCF-21 protein expression and CT vessel convergence were associated with pathological MVI in early-stage NSCLC. These preliminary findings suggest that an integrated molecular-radiological approach may support preoperative risk stratification after validation in larger external cohorts.