Pretreatment arterial spin labeling combined with radiological depth of invasion predicts treatment outcome in nonmetastatic NPC.

Abstract

Listen

Translate article

To investigate the value of arterial spin labeling (ASL) and radiological depth of invasion (rDOI) in predicting long-term treatment outcomes in nonmetastatic nasopharyngeal carcinoma (NPC). A total of 113 patients with NPC were included and randomly divided into training (n = 81) and validation (n = 32) cohorts. Tumor blood flow (TBF) parameters derived from the ASL (TBFMean, TBFSD, and nTBF) were obtained. Two radiologists independently measured the rDOI in both axial (rDOI_a) and coronal (rDOI_c) planes, subsequently categorizing patients into low-risk and high-risk groups. Spearman analysis was used to explore the correlation. In the training cohort, ASL-based, ASL+rDOI, and TNM models were constructed using univariate and multivariate Cox regression analyses. Model performance, including calibration, robustness, discrimination, and clinical utility, was assessed in both training and validation cohorts. The net classification index (NRI) and integrated discrimination improvement (IDI) were calculated. The median follow-up duration was 63.4 (56.2, 100.1) months. Disease progression and death occurred in 45 (39.8%) and 32 (28.3%) patients, respectively. TBFMean and rDOI were independent predictors of overall survival (OS) (hazard ratios [HR], 0.973 and 2.975, respectively) and progression-free survival (PFS) (HR, 0.985 and 2.207, respectively). rDOI_a and rDOI_c were significantly correlated with T stage (r = 0.538-0.738, P ≤ 0.002). The ASL+rDOI model demonstrated good robustness and clinical utility in both training and validation cohorts. Post-hoc subgroup analysis showed favorable results. TBFMean and rDOI can predict survival outcomes in patients with NPC. Adding rDOI further improved the prediction performance.