In our retrospective study involving 836 studies of gadoxetic acid–enhanced MRI, transient severe motion (TSM) artifacts occurred in 5.14% of examinations. This prevalence is consistent with previously reported in Asian cohorts, such as in a large Korean study involving 2230 patients by Jang et al. [11] which revealed a 5.0% prevalence Additionally, our findings are consistent with the subgroup analysis of Asian studies in the 2022 meta-analysis by Kim et al. [1], which reported a pooled TSM incidence of 8.8%. In contrast, Western studies in the same meta-analysis demonstrated a significantly higher pooled incidence of 16%, suggesting possible ethnic, anatomical (such as higher BMI in western population), procedural, or technical variations contributing to the discrepancy. In addition, our study found that older age (≥ 65-year-old) and lower serum albumin levels were identified as significant independent predictors of TSM. History of DM showed association with a higher risk of TSM in univariable analysis but not an independent predictor in multivariable analysis.

To date, the precise mechanism underlying gadoxetate disodium–mediated transient severe respiratory motion artifact (TSM) remains unclear. Several hypotheses have been proposed [22], including bronchospasm, central nervous system (CNS)-mediated tachypnea, and cardiac-mediated tachypnea. However, our current dataset does not allow us to thoroughly evaluate these specific mechanisms. Notably, the previously observed dose-dependent relationship between administered contrast volume and the incidence of TSM supports a physiological, rather than an allergic-like, mechanism [22, 23].

The association between older age and increased risk of TSM observed in our study aligns with findings from prior investigations [11, 12] that have consistently reported age as a significant factor influencing artifact severity. Multivariable analysis also demonstrated older age of more than 65-year-old to be independent predictors of TSM occurrence (OR = 4.59). This relationship could be accounted from reduced physiological reserve and decreased ability to maintain stable breath-holds among older patients, potentially due to cognitive decline, diminished respiratory function, or reduced muscle strength.

In addition to older age, lower serum albumin level was also demonstrated as a significant independent risk factor for TSM in our study. Our findings align with prior research that identifies low serum albumin as a potential risk factor for TSM during gadoxetate disodium-enhanced MRI. Specifically, Tamura et al. (2022) [24] demonstrated that hypoalbuminemia was independently associated with increased TSM occurrence in a cohort of 51 patients, with statistical significance maintained in multivariate analysis (P = 0.035). This study was further supported by additional animal data performed in the same study showing that albumin-diluted gadoxetate disodium delayed and reduced contrast-induced tachypnea, highlighting a mechanistic role for albumin in modulating respiratory stability during imaging.

The identification of hypoalbuminemia as an independent predictor of TSM is biologically plausible and clinically relevant. Serum albumin serves as a composite marker of nutritional status, hepatic synthetic function, and systemic health. Patients with low albumin may have reduced physiological reserves, impaired respiratory mechanics, or increased susceptibility to procedural stress, all of which could compromise their ability to maintain adequate breath-holding during the critical arterial phase of imaging. Moreover, some evidences have shown the physiological role of albumin in binding and transporting various substances, including pharmaceutical agents, provides a plausible mechanistic link to the observed association. Albumin serves as a major plasma carrier protein, modulating the pharmacokinetics of many endogenous and exogenous compounds [25]. Although gadoxetate disodium is only modestly protein-bound [26], it is reasonable to hypothesize that hypoalbuminemia could alter the distribution, plasma concentration, or clearance of the contrast agent. Such changes might amplify transient physiological responses, such as tachypnea or impaired breath-hold capacity, by increasing the free (unbound) fraction of the agent reaching systemic or central nervous system targets. Together, these findings suggest that serum albumin may serve as both a general health marker and a possible mediator of altered contrast-agent pharmacodynamics, contributing to transient respiratory dysfunction during hepatobiliary imaging. Further prospective studies with standardized imaging protocols are warranted to validate albumin’s role and determine its predictive utility for TSM risk stratification.

It is also important to acknowledge that some investigations have not consistently replicated these findings. Notably, Jang et al. [11], in a larger cohort of 2,230 patients, found that although hypoalbuminemia was associated with TSM on univariate analysis, it did not remain a significant independent predictor in multivariate models. This discrepancy may reflect differences in patient selection, MRI protocols, or analytical methods. It also suggests that while hypoalbuminemia may contribute to TSM risk, it may interact with other physiological and procedural factors, requiring a more nuanced, multifactorial assessment.

Contrary to several prior studies [7, 11, 12, 23], we did not observe a significant association between TSM and sex, BMI, DM, COPD, previous TSM, or fluid overload states such as pleural effusion and ascites. This discrepancy may be attributed to multiple factors. First, the prevalence of obesity and COPD in our study population was relatively low compared to Western populations, potentially limiting the statistical power to detect such associations. Second, the imaging protocol used at our institution, including consistent breath-hold coaching and standardized timing, may have mitigated the impact of these variables. Third, it is possible that patients with significant fluid overload were triaged to alternative imaging modalities such as contrast-enhanced CT, thus limiting their representation in our sample. Also, some studies have not identified any significant risk factor [15,16,17,18] and some reported using only univariable analysis.

Interestingly, factors previously reported as protective against TSM, such as prior MRI experience and HBV infection [11], were not found to be statistically significant in our cohort. While these findings have been interpreted in the past as markers of familiarity or enhanced physiological stability, our results suggest that such effects may be modest or population specific. Cultural, linguistic, or institutional differences in patient preparation may also explain the variability observed across studies.

For clinical implications, our findings showed importance for both clinical practice and imaging protocol design. The strong and independent association between older age and hypoalbuminemia and TSM suggests that these variables especially low serum albumin, could be incorporated into pre-imaging risk assessment protocols. Patients identified as high-risk based on low albumin levels might benefit from extended breath-hold coaching, the use of motion-resistant sequences, or alternative contrast agents. For example, multiple arterial phase acquisitions, compressed sensing protocols, or free-breathing arterial phases may offer viable solutions in such populations.

In addition, knowledge of hypoalbuminemia prior to imaging may help radiologists anticipate the potential for motion-related artifacts and guide post-processing or repeat scanning strategies when necessary. From a broader perspective, our findings also support a more personalized approach to liver MRI, in which patient-level risk factors guide imaging technique and contrast selection.

This study possesses several strengths. The relatively large cohort size and uniform imaging protocol. The use of stepwise logistic regression for variable selection ensured model parsimony while accounting for potential multicollinearity among predictors. Moreover, the study adds to a growing body of literature by highlighting the clinical relevance of serum albumin, a routinely measured and modifiable parameter, as a risk factor for motion artifacts in hepatic MRI.

However, certain limitations should be acknowledged. The retrospective design inherently limits causal inference and is subject to selection bias. Additionally, we did not account for psychological factors such as anxiety, claustrophobia, or sedation use, which may influence breath-hold performance. Technical parameters such as bolus timing, injection rate variability, and coil positioning were not varied and thus not assessed in our protocol. Additionally, that manual injection lacks the precision of a power injector, and that operator variability may have influenced TSM occurrence in our protocol. Although consensus grading reduces inter-observer variability, it may introduce bias when experience levels differ between readers, potentially allowing one opinion to dominate the final assessment. The number of patients with severe pleural effusion or ascites was relatively small, which may have limited our ability to detect associations for these variables. Lastly, although the logistic model performed well overall, convergence issues due to sparse data for rare binary predictors highlight the need for larger multicenter datasets.

Future research should aim to validate these findings in prospective, multicenter settings, ideally including broader geographic and demographic diversity. In particular, the relationship between serum albumin and TSM should be further explored in patients with advanced liver disease, where hypoalbuminemia is common and motion artifacts can significantly impair lesion detection.

Considering our findings, serum albumin, a component of several established liver function scoring systems such as Child-Pugh and Model for End-Stage Liver Disease (MELD), plays a critical role in assessing hepatic reserve. Given its strong association with TSM in our study, we propose that future research should incorporate composite scores like Child-Pugh or MELD to better evaluate liver functional status and its potential impact on motion artifact susceptibility.

Additionally, studies designed to evaluate technical modifications such as motion-robust sequence design, breath-hold training interventions, and body-weight-adjusted contrast injection may offer new strategies for reducing artifact prevalence in high-risk patients.

From a translational standpoint, artificial intelligence (AI) may also play a role in the future by predicting TSM risk in real-time and adjusting acquisition parameters accordingly. Finally, exploring correlations between serum albumin and respiratory waveform patterns during MRI may help elucidate the physiological underpinnings of motion susceptibility and open new avenues for preventive imaging protocols.