Ferumoxytol-enhanced MR Venography of Central Veins: Commentary

See also the article by Gallo et al.

J. Paul Finn, MD, is professor of radiology, medicine, and biomedical physics at UCLA. He is a past president and fellow of the International Society for Magnetic Resonance in Medicine (ISMRM), a fellow of the American Institute for Medical and Biological Engineering (AIMBE), and a board member of the Society for Cardiovascular Magnetic Resonance (SCMR). His research interests include MRI in congenital heart disease and the evolving use of ferumoxytol for cardiovascular MRI.

In this issue of Radiology: Cardiothoracic Imaging, Gallo et al (1) describe their experience using ferumoxytol-enhanced MR venography (FE-MRV) in a cohort of 35 patients with impaired renal function suspected of having central thoracic venous disease. Using catheter venography as the reference standard, they found that FE-MRV had a sensitivity of 99% and a specificity of 98% for diagnosis of venous stenosis or occlusion. Whereas diagnostic accuracy of this degree may seem unusual, it has now been reported in multiple studies using ferumoxytol for MRV (2–4). Most patients reported in these studies have had severe renal failure, where concerns persist about the use of gadolinium-based contrast agents (GBCAs). Although the macrocyclic GBCAs are felt to pose only a low risk of nephrogenic systemic fibrosis, the observation of gadolinium deposition in brain and bone (5) has generated unease about the long-term safety of GBCAs, particularly with repeated or frequent use. Ferumoxytol is gaining momentum as a viable alternative to GBCAs for vascular applications, and there is mounting evidence of its near-ideal properties for venographic imaging.

The current study by Gallo et al (1) affirms the high diagnostic accuracy of FE-MRV in a large patient cohort with disease affecting the central thoracic veins. Stoumpos et al (2) used ferumoxytol for vascular mapping of the upper limbs and central veins in 59 candidates for dialysis fistulae and reported a high accuracy in predicting a successful surgical outcome. Shahrouki et al (3) and Luhar et al (4) found 100% sensitivity and 100% specificity values in adults and children with disease affecting the thoracic and abdominal veins. The consensus of these studies is that FE-MRV may play a crucial role in patients with complex disease who require definitive diagnosis, and it may become the new reference standard.

Ferumoxytol is an iron nanoparticle approved only in the United States for intravenous treatment of iron deficiency anemia at all levels of renal function. Originally designed as a vascular MRI agent, it was repurposed to the iron therapy market. Ferumoxytol has a T1 relaxivity four to five times that of GBCAs and, because of its particle size, it remains unaltered in the blood stream for several hours after injection. Venous imaging can therefore be carried out anytime within a wide time window, without the need for bolus timing.

Whereas most works performing FE-MRV have focused on the steady-state distribution of ferumoxytol, Gallo et al (1) used both dynamic imaging and steady-state imaging. They administered the ferumoxytol as a 15-second bolus of dilute contrast material (1:5 in normal saline), analogous to first pass MR angiography with GBCA, and they acquired dynamic three-dimensional data every 5 seconds for 3 minutes. They then acquired three-dimensional T1-weighted images in steady state and used both the dynamic and steady-state images to reach the diagnosis. It is not clear to what extent the dynamic and steady-state images complemented each other or whether they would have generated the same high diagnostic accuracy if interrogated independently. The results of other studies suggest that venous stenosis and occlusion are readily detectable using only the steady-state three-dimensional images, but it seems likely that the dynamic images would provide additional information about the sequence of enhancement of primary and collateral venous pathways. It is also noteworthy that Gallo et al used a (dilute) bolus injection technique in all 35 patients over a 6-year time span (including an additional 37 patients who were excluded from analysis for lack of correlative venography). They found no adverse events in any patients, all of whom were closely monitored. The home institution of Gallo et al is a major contributing partner to the FeraSafe international multicenter registry (6) where partner sites contributed safety data on the diagnostic use of ferumoxytol. The FeraSafe registry was formed following a Food and Drug Administration (FDA) black box warning issued in March 2015 that highlighted postmarketing reports of severe hypersensitivity reactions to fast bolus therapeutic injection of ferumoxytol (7). The FDA withdrew the original indication for fast bolus injection at 30 mg per second and recommended slow intravenous infusion of dilute ferumoxytol at 30 mg per minute. The bolus injection rate used by Gallo et al is less than half the rate originally approved by the FDA. For context, a 510 mg therapy dose in a 70-kg adult corresponds to 7.3 mg per kilogram of body weight. The diagnostic dose of 3 mg per kilogram of body weight used by Gallo et al corresponds to 41% of the therapy dose, administered over about the same time as the original bolus therapy dose. So, whereas the injection rate and total dose used by Gallo et al are lower by a factor of two or more relative to the original therapy scheme, the rate is still far higher than recommended in the updated FDA guidelines. The exact relationship between injection rate and rate of infusion reactions to ferumoxytol is not clear. Mounting evidence from the FeraSafe registry, as well as from ongoing diagnostic clinical work, suggests that in closely monitored patients the rate of serious adverse events is so low as to be not yet defined. Early results from the FeraSafe registry reported zero serious adverse events in more than 4000 diagnostic injections, and the rate of minor adverse reactions was lower than reported in the package inserts for the macrocyclic GBCA (6). Whereas these safety data are reassuring, default infusion rates for ferumoxytol should follow the updated FDA guidelines (7), as Gallo et al state in their article. For specific clinical questions, it may be necessary to inject at a faster rate (albeit with diluted contrast material and in small doses), and these should be considered carefully on a case-by-case basis, balancing anticipated benefit and risk.

Most authors preface their articles on FE-MRV with a nod to the high diagnostic accuracy previously reported using gadolinium-enhanced MRV with extracellular GBCA. However, early enthusiasm for extracellular GBCA gave way to even more enthusiasm for the blood pool gadolinium agent gadofosveset (Ablavar; Lantheus Medical Imaging, North Billerica, Mass). It is well recognized that MRV with extracellular GBCA is sensitive to timing issues, and study quality can be variable. Further, prior to the nephrogenic systemic fibrosis era, MR venography was commonly performed with triple-dose GBCA (8) due to dilution effects, but this approach is no longer viable. Gadofosveset underwent reversible binding to serum albumin, which endowed it both with a longer intravascular residence time for venous imaging and a high T1 relaxivity. These properties made it a superior vascular imaging agent that could also be used in lower dose than the extracellular agents with which it shared a renal elimination pathway. Gadofosveset was withdrawn from production for commercial reasons in 2017, creating an unanticipated vacuum in the blood pool agent inventory that ferumoxytol now fills. Formal comparisons between gadofosveset and ferumoxytol for MRV are sparce, but limited reports suggest that ferumoxytol is at least equivalent (9). Whereas ferumoxytol can be regarded as a pure blood pool agent, gadofosveset lies somewhere in the space between a blood pool agent and an extracellular agent, with a variable plasma half-life that depends on protein binding and that is, in all cases, shorter than that of ferumoxytol.

Several investigators have pointed out that, whereas catheter venography is the conventional reference standard against which comparative techniques are measured, it has major limitations. Injection into a major artery (a source) is expected to opacify all patent branches downstream, but injection into a vein (a sink) is expected to opacify only the most direct pathway to the heart, whether normal or collateralized. Smaller local venous tributaries will not be routinely opacified (unless retrogradely by high-pressure injection), and separate injections are required for veins not on the primary pathway. Catheter venography, therefore, provides information piecewise and with limited scope. The ability of a blood pool agent to opacify all patent veins, independently of flow pathways, is clearly a huge advantage, but also casts doubt on the validity of catheter venography as a reference standard. The sensitivity of catheter venography for detection of disease in vessels not injected cannot be expected to be high!

In multiple studies, the imaging community has clearly established the need for an agent with ferumoxytol-like properties that is widely available, can be used dynamically and in steady state, has an imaging indication, and has an affordable price tag. By means of off-label use, ferumoxytol has been shown to meet many of these criteria, but it is not yet widely available, and it carries a high price tag. These limitations are ultimately addressable by industry, with the approval of federal regulators. The sooner this pathway is pursued for ferumoxytol itself or for an agent with ferumoxytol-like properties, the more patients stand to benefit from its established applications and its as-yet unexplored potential.