Improved Quality and Diagnostic Confidence Achieved by Use of Dose-Reduced Gadolinium Blood-Pool Agents for Time-Resolved Intracranial MR Angiography

Discussion

The present study confirms the feasibility and qualitative superiority of the BPA gadofosveset disodium over the SCA gadobenate dimeglumine for use in time-resolved cerebral MRA. The findings suggest that several theoretic advantages to the use of BPA for such CNS applications may be fully realized when dosing strategies and delivery are optimized. Specifically, the higher protein-binding capacity of BPA (80–96% reversible albumin binding), together with augmented T1 relaxation enhancement at clinical field strengths (r1 = 19 versus r1 = 6.3 at 1.5 T) increases the duration of the available acquisition window and provides improved vascular-to-background contrast, respectively.⁹ Furthermore, the benefits of greater T1 relaxivity afford flexibility in dosing strategies, as demonstrated by the 70% dose reduction (0.03 mmol versus 0.1 mmol) in our implementation, notably at uniformly improved angiographic quality. Given the superiority of gadobenate over many other extracellular SCA, including its greater relaxivity and its transient protein interaction, we believe that our findings may be readily generalizable to many other SCA as well.^9,18,19

Although the benefits of BPA for use in high-resolution, steady-state MRA were the subject of earlier studies, more recent reports have highlighted their potential advantages for use in non-CNS TR-MRA.^{4,13⇓⇓⇓–17} A number of vendor-based approaches to accelerated time-resolved imaging have been introduced, generally coupling some combination of parallel acceleration and variable k-space sampling attenuation with temporal interpolation and view sharing.^1,3,20 While facilitating the timing of contrast-enhanced MRA and approximating catheter angiographic physiologic information, such approaches may be prone to various degrees of temporal smearing, more common with traditional keyhole undersampling algorithms, which may confound interpretation of dynamic information when severe. We have not observed these effects to be particularly problematic in most cases; however, practitioners should remain aware of their potential to mimic shunt physiology and premature venous enhancement.

We have adopted a slower injection rate for BPA than that used in a similar, recent study by Frydrychowicz et al¹³ comparing BPA with SCA in thoracic TR-MRA (1.5 mL/s versus 3 mL/s, respectively). In combination with other factors, this difference may be nontrivial, given the nearly uniform superiority of BPA over SCA in our study, in contrast to the qualitative comparability between the 2 agents as described by the authors therein. Whereas differences in thoracic versus cerebral anatomy, particularly differences in vessel size, make conclusive judgments in this respect difficult, the significance of injection rate is strongly considered, given the preferential effects of BPA on transverse (r2) over longitudinal (r1) relaxation enhancement, approaching 6:1 at 3T.⁹ This may be especially problematic at high field, or in large vessels such as the subclavian arteries, in which unanticipated signal loss may occur at peak bolus as the result of T2* effects. Dose and injection rate reductions, as well as lower field scanning, may therefore ameliorate such factors when relevant. The use of 1.5T clinical systems may have further benefited this investigation, as the r1 and the rate of longitudinal relaxation enhancement (R1) have shown to be optimized for the macromolecular structure of BPA at 1.5T, whereas the differences compared with SCA are less pronounced at 3T.⁹ In our experience, inner-thigh and midsection burning and discomfort, sometimes described on more rapid injection of BPA, are mitigated by the slower injection rate prescribed in our studies.²¹

Frydrychowicz et al¹³ estimated both SNR and contrast-to-noise ratio (CNR) between SCA and BPA; however as they point out, and as thoroughly expounded previously by Reeder et al,²² SNR and CNR measurements of multichannel receive datasets are not straightforward, given the spatial nonuniformity of noise amplification inherent to parallel acceleration. For this reason, we believe qualitative estimates of image quality and quantitative descriptions of contrast kinetics to be more relevant to clinical practice. It is worth noting, however, that their study described the inferiority of arterial SNR and CNR with BPA, again possibly relating to degradation in imaging quality related to T2* effects. Interestingly, their findings described a reversal of this effect during the venous phase, in which relative dilution of the BPA is believed to restore the benefits of greater longitudinal relaxation enhancement. Among the quantitative details considered in this study, we found no significant difference in the likelihood of the 2 agents to produce uncontaminated (ie, free of venous enhancement) arteriograms; however, as might be anticipated from the slower injection rate of BPA, the peak arterial phase was encountered significantly later than with SCA.

Although use of early formulations of blood-pool prototypes was complicated by their tendency for tissue retention, gadofosveset trisodium was the first BPA, which progressed to human trials because of its efficient excretion.¹¹ Its high protein binding occurs almost immediately on injection, ensuring prolonged plasma half-life without extravasation—and therefore excellent blood-to-tissue contrast—while the maintenance of a predictable equilibrium between free and bound fractions ensures ready glomerular filtration of the free fraction, thus preventing retention within viscera.¹¹ Together with the excellent relaxation enhancement properties of its macromolecular structure, the profile of gadofosveset is therefore well-suited to such applications.

Limitations of this study include its retrospective design; however, both readers were fully blinded to the administered agent before review of both the raw volumes and MIP datasets. We chose to perform consensus reads of the study data to ensure uniformity in qualitative analyses, because our primary aim was to establish the qualitative superiority of an administered agent for TR-MRA rather than specifically to investigate the comparative diagnostic accuracy of one agent against the other. Given the extremely large datasets, which were examined across broad vascular anatomic boundaries precluding reliable uniformity in data presentation, we found a consensus evaluation of angiographic quality among the proposed metrics to be most generalizable to future clinical use. This approach, unfortunately, did not permit determination of interreader variability. All data were, however, reviewed with a uniform and reproducible methodology to approximate clinical practice as closely as achievable.

Our large study population of 100 TR-MR angiographies, comprising 50 studies, each performed with either BPA or SCA, constitutes a considerably larger study sample than comparable prior investigations, ranging between 10–30 TR-MRA examinations.^{4,13,15⇓–17,23⇓⇓–26} Not all patients were disease-free in our study population; attention was focused on the anatomically normal side for analysis, and we thus believe the findings to be readily generalizable to clinical practice and advocate strongly for the use of BPA for cerebral TR-MRA when feasible. We propose that the superior quality, comparable enhancement kinetics, and flexibility toward a 70% dose reduction compared with SCA underscore the advantages of BPA for TR-MRA, particularly at 1.5T scanning; however, at approximately 4 times the cost per dose, supplanting standard agents for all applications may be impractical at present.