Respiratory-Triggered Flow-Independent Noncontrast Non-ECG-Gated MRV (REACT) Versus Contrast-Enhanced MRV for Central Venous Evaluation in Children and Young Adults: A Six-Reader Study.

Please see the Editorial Comment by Abhay Srinivasan discussing this article. Contrast-enhanced MRI is commonly used to evaluate thoracic central venous patency in children and young adults. A flow-independent, noncontrast non-ECG-gated 3D MR angiography-MR venography (MRV) technique described in 2019 ["relaxation-enhanced angiography without contrast and triggering" (REACT)] may facilitate such evaluation. To compare image quality, diagnostic confidence, and interreader agreement between respiratory-triggered REACT and 3D Dixon-based contrast-enhanced MRV (CE-MRV) for evaluating thoracic central venous patency in children and young adults. This retrospective study included 42 consecutive children and young adults who underwent MRI of the neck and chest to evaluate central venous patency between August 2019 and January 2021 (median age, 5.2 years; IQR: 1.4-15.1 years; 22 female, 20 male). Examinations included respiratory-triggered REACT and navigator-gated CE-MRV sequences, based on the institution's standard-of-care protocol. Six pediatric radiologists from four different institutions independently reviewed REACT and CE-MRV sequences, assessing overall image quality (1-5; 5=excellent), diagnostic confidence (1-5; 5=extremely confident), and presence of clinically relevant artifact(s). Readers classified seven major central vessels as normal or abnormal (e.g., narrowing, thrombosis, or occlusion). Analysis used Wilcoxon signed-rank and McNemar tests, and Fleiss' kappa coefficients. Distribution of overall image quality scores was higher (p=.02) for REACT than for CE-MRV for one reader (both sequences: median score, 5). Image quality was not significantly different between sequences for the remaining five readers (all p>.05). Diagnostic confidence and frequency of clinically relevant artifact(s) were not significantly different between sequences for any reader (all p>.05). Interreader agreement for vessel classification (normal vs abnormal) was similar between sequences for all seven vessels (REACT, κ=0.37-0.81; CE-MRV, κ=0.34-0.81). Pooling readers and vessels, 65.4% of vessels were normal by both sequences, 18.7% abnormal by both sequences. 9.8% abnormal by REACT only, and 6.1% abnormal by CE-MRV only. Respiratory-triggered REACT, in comparison with CE-MRV, demonstrated no significant difference in image quality (aside from for one of six readers), diagnostic confidence, or frequency of artifact(s), with similar interreader agreement for vessel classification as normal or abnormal. High-resolution 3D MR venography performed without IV contrast material can be used to assess central venous patency in children and young adults.