Lesion Visualization of an Oral Manganese Contrast Agent Compared to Unenhanced MRI and Gadobenate Dimeglumine in Patients Undergoing Liver Magnetic Resonance Imaging for Evaluation of Colorectal Cancer Metastases: Centralized Assessment of a Randomized, Crossover, Phase II Study.

MATERIALS AND METHODS

Study Design, Population, and Exposures
The present study was a centralized assessment with independent blinded readers conducted from July to November 2020, analyzing MRI of 20 patients from a previously completed study (Fig. 1),21 which was performed at Karolinska University Hospital Huddinge in Stockholm, Sweden, from November 2005 to September 2007. The original study was an open-label, randomized, crossover study in 20 patients with known or suspected colorectal liver metastases.
Briefly, the original study prospectively and consecutively recruited 20 adult patients, including 10 men and 10 women aged 51–79 years. All of them had been referred to the multidisciplinary team for evaluation of known or suspected liver metastases, before partial liver resection. In the study, patients were randomized to receive either ACE-MBCA followed by gadobenate dimeglumine (n = 11) or gadobenate dimeglumine followed by ACE-MBCA (n = 9) in random order on the basis of sealed envelopes, with MRI scans performed before and after the administration of each contrast agent. Liver MRI was performed at 1.5 T via a combination of spine and flexible body array coils, with ACE-MBCA MRI conducted before and 3 hours after the oral administration of 1.6 g of MnCl2
18 (dissolved in water with alanine and vitamin D3), whereas gadobenate dimeglumine followed a 5-phase protocol after IV administration of 0.1 mmol/kg. The protocol included 3D T1-weighted volumetric interpolated breath-hold examination (VIBE) and T2-HASTE sequences, with total examination times of approximately 20 minutes for ACE-MBCA sessions compared with 70 minutes for the comprehensive gadobenate dimeglumine protocol. For the calculations after contrast administration, only the hepatospecific phases were used (i.e. the T1-weighted image 3 hours after ingestion of ACE-MBCA or the T1-weighted image 2 hours after IV of gadobenate dimeglumine, respectively) (Supplementary Table 1, http://links.lww.com/RLI/B9).21

Evaluations
Three independent radiologists, with 15, 30 and 35 years of experience, who were blinded to the contrast agent, evaluated different types of assessments in sequential sessions with at least 1 week interval between assessments of the same patient to reduce recall bias. The evaluations included unenhanced MRI sequences acquired before contrast agent administration, contrast-enhanced MRI sequences obtained after oral ACE-MBCA and after IV gadobenate dimeglumine administration, and combined MRI evaluations where unenhanced and contrast-enhanced sequences were evaluated together, performed separately for ACE-MBCA and gadobenate dimeglumine, resulting in ACE-MBCA combined and gadobenate dimeglumine combined, respectively. The dynamic phase after the IV of gadobenate dimeglumine was not used in this analysis as was the case for the original data collection (Supplementary table 1, http://links.lww.com/RLI/B9). Reads for unenhanced and gadobenate dimeglumine enhanced MRI were not in scope for the purpose of investigating its efficacy. During each session, readers marked up to 15 lesions per patient by placing an arrow in the center of each lesion on the slice showing the maximum diameter. For each lesion, readers recorded the liver segmental location and scored lesion border delineation and lesion contrast using 4-point qualitative scales (1 = poor, 2 = partial/moderate, 3 = good, 4 = excellent). After the completion of all the primary reader assessments, a fourth independent reader evaluated all the images to identify and match the lesions that were visible across the individual read sessions, tracking the lesions detected by each of the 3 primary readers across unenhanced, enhanced, and combined MRI sessions for both contrast agents. This fourth reader then performed additional quantitative assessments of the matched lesions.
The primary outcome measured changes in lesion border delineation (LBD) and lesion contrast (LC) scores for up to 15 lesions per patient.21 Additional outcomes included comparative visualization assessments between different MRI combinations, lesion detection, lesion size measurements, reader confidence, and quantitative contrast parameters. These measurements were done by all 3 readers.
The lesion size was measured as the longest diameter of the smallest and largest lesions per patient. Confidence in lesion detection and localization was rated on a 3-point scale (low, moderate, high). Quantitative parameters were assessed by reader 4, measuring the mean percent signal intensity enhancement of liver (SI%), mean liver-to-lesion contrast (LLC),22 mean signal-to-noise ratio (SNR), and mean contrast-to-noise ratio (CNR) of up to 5 lesions per patient within each image set. Derived parameters were computed as:
SI% = ([SIliv post contrast – SIliv pre contrast]/[SIliv pre contrast]) × 100

LLC = (SIliv – mean of SIles)/(SIliv + mean of SIles)

SNR = SIliv/SDnoise

CNR = (SIliv – mean SIles)/SDnoise

SIliv = signal intensity of the liver; SIles = signal intensity of the lesion; SDnoise = standard deviation of the background noise of air immediately outside the patient; mean of SIles = mean of all lesion signal intensities for a patient.
Overall image quality was rated as optimal, suboptimal, or not evaluable.

Data Analysis
To maintain the blinded nature of the study, all statistical analyses were performed by an independent statistician who was not involved in the image reading process, using SAS platform v9.4. The primary analysis evaluated the mean changes in lesion visualization (LBD and LC) for each patient between unenhanced MRI and ACE-MBCA combined MRI (Fig. 1) via 1-sided paired t tests. The superior scoring of combined MRI over unenhanced MRI for visualization was evaluated through LBD and LC scores from 3 independent readers (Supplementary data, http://links.lww.com/RLI/B9). Success required 2 out of 3 readers to demonstrate statistically significant improvements in both parameters at a 1-sided significance level of 0.025.
The secondary efficacy analyses included evaluating changes in visualization for ACE-MBCA enhanced MRI versus unenhanced MRI and ACE-MBCA combined MRI versus gadobenate dimeglumine combined MRI (Fig. 1), using the same superior scoring criteria as the primary analysis, as described here.
For comparative analyses, descriptive statistics, including arithmetic means, standard deviations, medians, ranges, and 95% confidence intervals, were generally provided. No formal hypothesis testing was performed for most secondary endpoints, but the 95% CIs were used to evaluate potential differences between categories.

Ethics Approval
The original study was conducted in accordance with its protocol, regulatory requirements, good clinical practice, and the Declaration of Helsinki and its amendments in force at the time of study initiation. As this study involved only a retrospective analysis of existing MRI data, with no new patient involvement or data collection, it did not require separate ethics committee approval. The present analysis was within the scope of what participants had originally consented to during the initial study enrollment; therefore, no additional consent was needed.

RESULTS

Study Population
The demographic and anthropometric characteristics of the studied population were comparable between groups and are presented in Table 1 and can also be found in the initial study publication.21

Visualization of Lesions
The results of the primary analysis between unenhanced MRI, ACE-MBCA enhanced MRI, and ACE-MBCA combined MRI are presented in Supplementary Table 2, http://links.lww.com/RLI/B9. The mean LBD scores for unenhanced MRI ranged from 1.8 to 2.3 across readers, whereas ACE-MBCA combined MRI scores ranged from 2.4 to 2.9. Similarly, the mean LC score increased from 1.8–2.3 for the unenhanced MRI to 2.8–3.3 for the ACE-MBCA combined MRI (Supplementary Table 2, http://links.lww.com/RLI/B9). The paired differences between ACE-MBCA combined MRI and unenhanced MRI were statistically significant in favor of the ACE-MBCA combined for both LBD and LC across all 3 independent readers (mean differences ranging between 0.6 and 1.5 for LBD and 0.9 and 2.2 for LC), demonstrating the superior scoring of ACE-MBCA combined MRI over unenhanced MRI for the visualization of matched focal liver lesions (Table 2).
As described in Table 2, LBD and LC scores increased between unenhanced MRI and ACE-MBCA enhanced MRI for readers 1 and 3 (1.1 and 1.6 for LBD and 1.2 and 2.0 for LC, all with P value <0.001) but not for reader 2 (−0.3 and 0.12, respectively, both with P value >0.2). A typical example of the contrast enhancement effect at MRI is illustrated in Figure 2, showing an axial T1-weighted image of the liver before and after the ingestion of ACE-MBCA.
As secondary result, for ACE-MBCA combined MRI versus gadobenate dimeglumine combined MRI, statistically significant differences were observed only for reader 1. No statistically significant differences were found for readers 2 and 3 in either LBD or LC, as described in Table 2.

Lesion Detection
All 3 readers detected significantly more lesions with both ACE-MBCA enhanced MRI and ACE-MBCA combined MRI than with unenhanced MRI (Supplementary Table 3, http://links.lww.com/RLI/B9), with positive 95% confidence intervals indicating statistically significant improvements. The increase in lesion detection with ACE-MBCA unenhanced MRI ranged from 0.5 to 1.4 additional lesions per reader, whereas ACE-MBCA combined MRI showed similar improvements, with 0.4 to 0.8 additional lesions detected per reader (Supplementary Table 3, http://links.lww.com/RLI/B9). When comparing ACE-MBCA combined with MRI to gadobenate dimeglumine combined MRI, there were no statistically significant differences in lesion detection. The numbers of lesions detected for ACE-MBCA and gadobenate dimeglumine by each reader are described in Supplementary Table 4, http://links.lww.com/RLI/B9.

Lesion Size
The results of the readers' measurements of lesion dimensions with both contrast agents are summarized in Supplementary Table 5, http://links.lww.com/RLI/B9. The smallest measurable lesions, ranging from 3 to 4 mm in size, were detected on both gadobenate dimeglumine MRI and ACE-MBCA MRI examinations (alone or in combination). Lesion measurements by all 3 readers showed larger mean longest diameters for ACE-MBCA MRI (enhanced and combined MRI) compared with unenhanced MRI (Supplementary Table 6, http://links.lww.com/RLI/B9). The comparisons between ACE-MBCA and gadobenate dimeglumine combined MRIs showed reader-dependent variations (Supplementary Table 6, http://links.lww.com/RLI/B9).
For smallest detected lesions, readers 2 and 3 measured smaller mean longest diameters with ACE-MBCA (enhanced and combined) MRIs than for unenhanced MRI but the differences were not statistically different (Supplementary Table 6, http://links.lww.com/RLI/B9).
For lesions <1 cm, ACE-MBCA combined MRI improved detection compared with unenhanced MRI, while showing similar detection rates to gadobenate dimeglumine combined MRI. In the 1–3-cm range, both ACE-MBCA enhanced and combined MRIs maintained detection rates comparable to unenhanced MRI, with ACE-MBCA enhanced MRI showing slightly better performance (Supplementary Table 7, http://links.lww.com/RLI/B9).

Confidence in Lesion Detection and Localization
The results of the readers' confidence in lesion detection with the ACE-MBCA showed variation across the 3 readers. For readers 1 and 3, the mean confidence scores were higher for both ACE-MBCA enhanced and combined MRIs than for unenhanced MRI and reader 2 showed a marginally lower mean confidence score for ACE-MBCA enhanced MRI than for unenhanced MRI and combined MRIs. In the case of gadobenate dimeglumine combined MRI, the readers' mean confidence scores for lesion detection were consistent across all the readers (Supplementary Table 8, http://links.lww.com/RLI/B9). No statistically significant differences in the readers' confidence in lesion detection were detected between ACE-MBCA and gadobenate dimeglumine combined MRIs for any of the 3 independent readers (Supplementary Table 9, http://links.lww.com/RLI/B9).
For lesion localization reader 1 and 3 reported greater confidence with both ACE-MBCA enhanced MRI and ACE-MBCA combined MRI than with unenhanced MRI, with statistically significant differences. Reader 2 showed slightly lower confidence with ACE-MBCA combined MRI compared with unenhanced MRI (nonsignificant) and similar confidence with ACE-MBCA enhanced MRI (Supplementary Table 10, http://links.lww.com/RLI/B9). When comparing ACE-MBCA to gadobenate dimeglumine combined MRIs, there were no statistically significant differences in confidence for any of the 3 readers (Supplementary Table 11, http://links.lww.com/RLI/B9).

Quantitative Parameters
The quantitative assessments are described in Supplementary Table 12, http://links.lww.com/RLI/B9. The mean LLC was greater with ACE-MBCA enhanced MRI than with unenhanced MRI (0.09, 95% CI 0.03–0.15), as was the CNR (1.51, 95% CI 0.53–2. 50). The mean SNR was not significantly different between ACE-MBCA enhanced and unenhanced MRI.
For gadobenate dimeglumine enhanced MRI, the LLC showed a mean difference of −0.17 (95% CI −0.24 to −0.10), and the CNR showed a mean difference of −4.20 (95% CI −6.77 to −1.64), whereas for gadobenate dimeglumine enhanced MRI, the SNR showed a mean difference of 2.23 (95% CI −0.53 to 4.99) compared with unenhanced MRI (Supplementary Table 13, http://links.lww.com/RLI/B9).

DISCUSSION
In this blinded evaluation, ACE-MBCA enhanced MRI demonstrated statistically significant superior scoring over unenhanced MRI for visualizing focal liver lesions, with all 3 independent readers showing improved lesion LBD and LC. This superiority was consistently supported across multiple assessment parameters, including improved lesion detection, enhanced measurement capabilities, and increased reader confidence. This study confirmed the ability of the ACE-MBCA to detect significantly more lesions than unenhanced MRI does, particularly for small lesions. Importantly, ACE-MBCA demonstrated comparable performance to gadobenate dimeglumine across all evaluation criteria, including lesion visualization, detection rates, size measurements, and reader confidence scores, although the study was not designed to formally test for statistical equivalence or noninferiority. The confidence score assessments for ACE-MBCA showed that this investigational contrast agent is a promising alternative, especially considering its oral dosing 2–3 hours before MRI and potential safety advantages for renally impaired patients.
The superior visualization of focal liver lesions achieved with ACE-MBCA combined MRI, demonstrated consistently across all 3 independent readers for both LBD and LC, has significant clinical implications for patient care. This enhanced visualization capability could enable more precise disease staging through better lesion detection, particularly for smaller lesions that might be missed on unenhanced MRI. The improved LBD could lead to more accurate surgical planning and better assessment of treatment response. Furthermore, the consistently superior lesion detection capability of ACE-MBCA, especially for small lesions, could facilitate earlier identification of liver metastases at a more treatable stage, or avoid futile liver resections in patients with widespread small metastases, making it particularly valuable for surveillance MRI in high-risk patients and monitoring treatment response. While 1 reader demonstrated statistically significant improvements in both LBD and LC with the ACE-MBCA compared with gadobenate dimeglumine in the hepatobiliary phase, the other 2 readers showed comparable performance between the agents in this phase, suggesting noninferior visualization capabilities. It is worth noting that gadobenate dimeglumine provides additional dynamic imaging phases (arterial, portal venous, and equilibrium) that were not available with ACE-MBCA, as the latter is an oral product that enters the liver via the portal vein. These improvements are especially significant for patients with renal impairment, who currently face limited MRI options owing to safety concerns for GBCA, potentially offering them access to high-quality contrast-enhanced MRI with a manganese-based alternative. The ability of ACE-MBCA to detect smaller lesions is also noteworthy. For lesions <1 cm, ACE-MBCA combined MRI improved detection compared with unenhanced MRI, while showing similar detection rates to gadobenate dimeglumine combined MRI.
In patients with severely impaired renal function, iodine-based contrast agents are typically contraindicated due to the increased risk of contrast-induced nephropathy. While recent advances in CT technology have enabled examinations with reduced doses of iodine-based contrast agents, the clinical utility and safety profile of these lower-dose protocols in this vulnerable patient population remain inadequately characterized.23–25 The safety profile and administration characteristics of the ACE-MBCA present some advantages for clinical practice, particularly for patients with renal impairment. The oral delivery route of ACE-MBCA utilizes physiological manganese transport and homeostatic mechanisms, resulting in 2 distinct advantages: optimized hepatic targeting via first-pass portal circulation, and regulated manganese absorption and metabolism through endogenous pathways. This pharmacokinetic profile contrasts with intravenous administration, as it enables controlled manganese biodistribution through normal physiological processes. The safety profile of the ACE-MBCA is characterized by only mild to moderate gastrointestinal events and no serious adverse effects, although 95% of participants experienced at least 1 adverse event, and 1 severe adverse event (back pain judged to be possibly related to ACE-MBCA) was reported. While the manganese dose is substantially higher than the recommended dietary intake, the safety profile has been further evaluated in subsequent clinical studies, including the recently completed SPARKLE trial.19,21 This finding suggests ACE-MBCA is a potentially viable contrast option for patients whose GBCAs are not recommended. Furthermore, the ACE-MBCA oral administration protocol offers practical advantages, allowing patients to self-administer the contrast agent 2–3 hours before the MRI based on established pharmacokinetic data from previous research26 and eliminating the need for IV injection, factors that could improve the overall patient experience and workflow efficiency in clinical practice.
The initial study21 with the present samples demonstrated that ACE-MBCA was as sensitive as an extensive IV gadobenate dimeglumine protocol in detecting colorectal liver metastases, with detection rates of 93% and 95%, respectively. While this initial study evaluated diagnostic performance metrics including false-positive rates, the current analysis specifically focused on standardized visualization parameters (lesion border delineation and lesion contrast) following a methodology recommended by regulatory agencies. To align with current regulatory standards and provide a more rigorous assessment of diagnostic performance, particularly for visualization endpoints, a blinded reanalysis was conducted using methodology that has become the standard approach for evaluating imaging endpoints in pivotal contrast agent trials, similar to recent studies such as SPARKLE19 and gadopiclenol.20 This standardized approach allows for direct comparison of visualization capabilities between contrast-enhanced and unenhanced MRI, while also enabling assessment of lesion detection patterns across different readers and MRI protocols.
This study presents several notable methodological strengths that increase the reliability of its findings, including the centralized blinded evaluation of MRI by 3 independent radiologists, which significantly reduced potential bias, and the comprehensive assessment of multiple imaging parameters, which provided a thorough evaluation of diagnostic performance. The use of separate independent readers, rather than a consensus read, reduced the impact of individual reader interpretation on the study conclusions, providing a more robust assessment of ACE-MBCA's visualization capabilities. The study's comparative design against both unenhanced MRI and gadobenate dimeglumine offered valuable insights into the relative performance of the ACE-MBCA compared with current clinical standards. However, several important limitations must be considered when interpreting these results: the small sample size of 20 patients limits statistical power and generalizability, and the retrospective nature of the analysis may introduce inherent biases such as observer bias—since readers are evaluating historical images with knowledge that these are from patients with confirmed liver metastases—or recall bias—as readers might still remember previous interpretations of images from the same patient across different sessions despite the between-read gaps. The open-label design of the study could have influenced the image acquisition process despite the blinded re-evaluation, and while this study compared ACE-MBCA to gadobenate dimeglumine, these results cannot be generalized to all gadolinium-based contrast agents, particularly since gadoxetic acid is now the preferred liver-specific GBCA in clinical practice but was not yet available when the primary study was designed. The interpretation of lesion size measurements should be approached with caution, as without pathological confirmation, it remains uncertain whether differences in lesion sizes between MRI methods represent true anatomical variations or measurement bias, as previous studies have shown that MRI measurements can both overestimate and underestimate actual lesion dimensions compared with pathological specimens. Additionally, diffusion-weighted imaging, which has become standard in abdominal protocols, was not available at the time of study planning. While this study was limited to adult patients with known liver metastases, future research should include long-term follow-up with direct clinical outcome measures to fully establish the utility of the ACE-MBCA in routine practice and define optimal type-specific diagnostic protocols for broader patient populations.

Supplementary Material

SUPPLEMENTARY MATERIAL