Venous mapping is important in end stage renal disease patients requiring hemodialysis for vascular access planning, to establish vessel caliber and patency¹. Ultrasound (US) is performed first line, but is difficult for central veins. Contrast-enhanced MRV (CE-MRV) is undesirable due to risk of Nephrogenic Systemic Fibrosis². Our purpose was to evaluate image quality (IQ) and measured vessel caliber of central and upper extremity veins with a non-contrast MRV (NC-MRV) protocol, with comparison to CE-MRV and US measurements in healthy volunteers.

Following informed consent, 10 healthy volunteers (3M, 7F, mean 36y) underwent bilateral NC-MRV and CE-MRV at 1.5T (Avanto, Siemens) in a single visit. NC-MRV comprised 2D and 3D bSSFP imaging of the chest, followed by 2-station fast spin echo MRV (FSE-MRV), based on a fresh blood imaging technique³. Upper FSE-MRV station (SVC to elbow) was performed with respiratory navigator in an oblique coronal plane, and lower station (forearm) in an oblique sagittal plane. Subsequently, 2 station CE-MRV images were acquired in identical orientation, 3 minutes post injection of 0.2 mmol/kg gadobenate dimeglumine (Multihance, Bracco) into a hand vein, with breath-holding for the upper station. Sequence parameters are provided in Table 1. Bilateral upper limb B-mode US from proximal arm to wrist was performed within 7 days of MR (Philips IU22, linear 15-7 array transducer), with measurements made by an experienced sonographer. A tourniquet was employed for lower MRV stations and US.

Anonymized MR images were independently evaluated by two vascular radiologists on a PACS workstation (Impax, Agfa). Up to 27 segments were evaluated, including central (SVC, brachiocephalic, subclavian and axillary) and superficial arm/forearm veins (basilic, cephalic, and medial antebrachial veins where present). IQ was scored on a 5-point scale (0=non-diagnostic, 2= sufficient for diagnosis, 4= excellent IQ). For both MR techniques and US, caliber was measured for veins larger than 3mm diameter. IQ and caliber measurements were combined across readers and compared with the Wilcoxon signed rank test.

All subjects successfully completed both MR and US. Overall, 471 segments were graded for IQ with NC-MRV and 468 with CE-MRV. Mean IQ was diagnostic for both sequences (Figures 1 and 2), but statistically significantly superior for CE-MRV versus NC-MRV (Table 2). CE-MRV was superior for central and forearm veins, with forearm segments suboptimally visualized (mean IQ score <2) with NC-MRV. Mean IQ was higher for NC-MRV for arm veins, approaching statistical significance. Difficulties encountered with both techniques were: assessment of the proximal humeral basilic vein related to patient positioning, and separating target veins from adjacent vessels in the upper limb. Vessel blurring was the main artifact reported limiting NC-MRV image quality, particularly for forearm segments, with flow dephasing in central segments in 2 subjects. Poor fat suppression and low SNR were cited as reasons for poor IQ for CE-MRV for distal segments. Caliber measurements were performed in 383 NC-MRV and 371 CE-MRV segments, with no overall significant difference in caliber between sequences. NC-MRV mean measurements were 0.1mm larger than CE-MRV in the arm, with no significant differences observed in central and forearm regions. A 0.6mm greater mean caliber for NC-MRV was observed for the distal humeral basilic vein, with no other significant segmental differences between NC-MRV and CE-MRV measurements (Table 3). 261 vessel segments were measured at US, with both NC-MRV and CE-MRV measurements significantly larger than US measurements (NC-MRV 8.5±5.2mm, CE-MRV 8.3±5.2mm, US 4.5±1.8mm, both p <0.001).

A non-contrast assessment of the upper extremity and central veins is desirable in patients with end stage renal disease for vascular access planning. Previous preliminary work demonstrated superior visualization of the arm veins with FSE-MRV in volunteers compared with 2D-TOF⁴. Similarly, superior venous IQ has been described using a balanced SSFP technique against CE-MRV⁵. In our study, IQ was inferior for NC-MRV for central and forearm veins. Vessel blurring in the phase encoding direction, related to interecho spacing and echo train length with FSE imaging⁶, was the major factor impacting IQ. Optimization with increased acceleration factors to minimize echo train length, in combination with 3T imaging, may engender improvements in image quality and spatial resolution.

We demonstrated comparable venous caliber of most segments for NC-MRV and CE-MRV, with small (<1mm) but significant differences observed in the basilic vein in the arm. Both MRV techniques yielded significantly larger vessel caliber than US measurements, as has been reported for bSSFP NC-MRV⁷, with direct US transducer pressure, patient positioning, temporal variation and MR spatial resolution limitations potential factors. These results are concerning if MRV is to be applied in clinical practice.