Our portal vein 4D flow MRI study examined rotational blood flow in 104 patients with cirrhosis. Lower portal venous vorticity was associated with high-risk esophageal varices. We evaluated these rotational features alongside conventional flow measurements and vessel geometry.
The study was published in Journal of Magnetic Resonance Imaging (2026). Read the original paper ↗
Why study rotational flow in the portal vein?
The portal vein receives blood from the splenic and superior mesenteric veins. Their confluence can create helical flow: blood moves forward while rotating around the vessel axis. Conventional flow rate and velocity describe how much blood moves and how fast, while rotational measurements describe aspects of its three-dimensional organization.
In cirrhosis, altered resistance and collateral pathways change portal circulation. We asked whether flow organization in the main portal vein was associated with the severity of esophageal varices, a complication of portal hypertension.
Study design and quantitative methods
This retrospective, single-center study included 104 patients who underwent liver MRI and endoscopy within three months of each other. Endoscopic findings defined three groups: no varices (n = 48), low-risk varices (n = 37), and high-risk varices (n = 19). High-risk status was based on larger varices or red-color signs, rather than a prospectively observed bleeding event.
| Measurement | What it describes | How it was assessed |
|---|---|---|
| Flow rate and mean velocity | Blood transport through the portal circulation | Retrospective planes in the main portal vein, splenic vein, and superior mesenteric vein |
| Vessel geometry | Effective diameter and confluence angles | Cross-sectional lumen area and vessel centerline vectors |
| Vorticity | Local rotation of the velocity field | Curl of velocity; intensity averaged over the main portal vein region |
| Helicity and h2 | Alignment of motion with rotation, and its overall intensity | Velocity–vorticity dot product; h2 averages absolute helicity over space and time |
| Localized normalized helicity | Alignment independent of velocity and vorticity magnitudes | Normalized velocity–vorticity dot product |
| Vortex volume | Volume occupied by rotation-dominated structures | Q-criterion threshold of 100 s−2, informed by a threshold sensitivity analysis |
Rotational measurements focused on the main portal vein between its proximal confluence and distal branching region. Magnitude and phase data underwent anti-aliasing, noise filtering, and eddy-current correction, followed by PC-MRA-based segmentation. Group comparisons used ANOVA with Tukey testing or Kruskal–Wallis with Holm-adjusted Conover comparisons. ROC analysis assessed discrimination of the high-risk group.
| Parameter | Protocol |
|---|---|
| Field strength and sequence | 3 T; three-directional velocity-encoded gradient echo; Cartesian sampling |
| TR / TE | 6.68 / 4.06 ms |
| Field of view / acquisition matrix | 292 × 360 mm² / 160 × 130 |
| Slice thickness | 2.5 mm; the paper describes an isotropic reconstructed velocity dataset |
| VENC | 30 / 20 / 20 cm/s in left–right / anterior–posterior / head–foot directions |
| Acceleration / gating | GRAPPA factor 3; prospective ECG and respiratory synchronization |
| Scan duration | 6–15 minutes |
Main finding: less rotational flow in the high-risk group
| Group | Participants | Vorticity, mean ± SD |
|---|---|---|
| No varices | 48 | 17.86 ± 3.26 s−1 |
| Low-risk varices | 37 | 16.65 ± 4.58 s−1 |
| High-risk varices | 19 | 13.59 ± 3.23 s−1 |
Vorticity differed across groups (P < 0.001), and the high-risk group had lower values than both other groups. Its ROC AUC was 0.77 (95% CI: 0.65–0.88). The study-derived threshold of ≤16.30 s−1 gave 78.9% sensitivity and 62.4% specificity. This is an exploratory threshold derived in this cohort, rather than an established clinical cutoff.
Helicity and h2 also declined with high-risk varices, with AUCs of 0.77 and 0.76. Their study-derived thresholds had higher specificity, approximately 87–89%, with sensitivity of 63.2%. Localized normalized helicity showed weaker discrimination (AUC 0.56), illustrating that these parameters capture different features of flow.
Mean portal vein velocity was lower in the high-risk group. Fractional portal flow change and confluence angles did not significantly differ across groups. The findings therefore suggest that rotational intensity provides a useful additional description of altered portal circulation, although incremental clinical benefit over existing assessment was not established.
What the findings mean
Lower rotational flow may reflect less organized portal hemodynamics as disease and collateral circulation progress. This is a physiological interpretation of the observed associations; the study does not establish that reduced vorticity causes varices or bleeding.
The next step is prospective validation against actual bleeding events, with standardized timing, fasting, and independent cohorts. Only 19 patients had high-risk varices, there was no healthy control group, and all data came from one center and MRI platform. The results support further investigation of rotational flow as an adjunct to current risk assessment.
Primary source
Park S, Moon CM, Kwon M, et al. Exploratory 4D Flow MRI Study of Portal Venous Rotational Flow for High-Risk Esophageal Varices in Cirrhosis. Journal of Magnetic Resonance Imaging. 2026;64:283–293. DOI: 10.1002/jmri.70287. Sungho Park and Chung Man Moon contributed equally.