Acute adverse events in cardiac MR imaging with gadolinium-based contrast agents: results from the European Society of Cardiovascular Radiology (ESCR) MRCT Registry in 72,839 patients

Gadolinium-based contrast agents (GBCAs) are considered safe in magnetic resonance (MR) imaging with acute adverse event rates reported to be ranging from 0.04 to 2.2% [1‐7].

Differences in adverse event rates by anatomical region might originate not only from distinct imaging algorithms but also from different propensities depending on the patient’s underlying pathology. In this context, cardiac imaging is of special interest, with a variety of indications ranging from cardiac viability assessment in older multimorbid patients to myocarditis imaging in primarily younger and healthier patients [8‐10].

Moreover, cardiac MR (CMR) imaging for ischemic heart disease is routinely performed with pharmacological stressors, which might increase the incidence of acute adverse events in general, and GBCA-related acute adverse events in specific [11].

So far, acute adverse events in CMR imaging have been systematically evaluated in the European Cardiovascular Magnetic Resonance (EuroCMR) Registry [12]. Adverse event rates ranged from 0.17% in a sample of 17,767 patients to 0.12% in a sample of 37,788 patients [5, 13]. However, there is no comprehensive literature evaluating acute adverse events in GBCA-enhanced cardiac MR imaging with statistical adjustment for potential confounders, such as stress test imaging, main indications, age, and sex.

Therefore, the aim of the current study was to evaluate data originating from the multicenter, multinational cardiac MRCT Registry of the European Society of Cardiovascular Radiology (ESCR) to assess the likelihood of gadolinium-associated acute adverse events in cardiac MR imaging.

This retrospective study was performed in accordance with the Declaration of Helsinki and received institutional review board approval (Leipzig University, No. 131/17-ek).

The data source of this study is the multinational, multicenter ESCR MRCT Registry, which includes imaging studies submitted between 2013 and 2016. Using a standardized online questionnaire, the physicians responsible for CMR imaging prospectively provided mandatory information on patient characteristics, indications, diagnoses, imaging techniques, contrast media application, and occurrence of acute adverse events (reported as the most severe event for each patient). Information on GBCA concentration and volume was non-mandatory. MRCT Registry users were unaware that data was utilized to assess acute adverse event incidences, thereby minimizing potential reporting biases.

Only CMR scans with intravenous administration of GBCA were included. GBCA molecular structure was classified as cyclic or linear, its ionic properties as ionic or non-ionic, and its thermodynamic chelate stability by log K_therm [14]. Covariates included pharmacological stressors, main indication, GBCA volume and concentration, gender, and age. Imaging indications rather than diagnoses were evaluated to avoid reverse causation, since acute adverse events might have influenced diagnoses (i.e., aborted examination or artifacts as a consequence of acute adverse events). Five administrations of gadoversetamide (Optimark^®, Medtronic-Covidien) were excluded from analyses to avoid non-convergence of the statistical models due to a too small subgroup size.

The overall rate of acute adverse events (AAEs) in our multinational, multicenter cohort was 0.36%. This finding is in line with the literature on GBCA application in general and cardiac imaging in specific, ranging from 0.04 to 2.2% [1‐7, 13, 18‐21].

However, AAEs were more frequent compared to the EuroCMR Registry in 2011 and 2015, reporting AAE rates of 0.12–0.17%, while using the same ACR classification.

In particular, the number of moderate and severe AAEs was higher in the ESCR MRCT Registry: 40%, 50.8%, and 9.2% of the reported acute adverse events were mild, moderate, and severe, compared to 17.2% moderate and up to 6.3% severe adverse events in the EuroCMR Registry [6]. However, the latter study did not explicitly include cardiac stress test imaging. Differences in AAE rates could be further attributed to multiple factors, including diverging sample sizes and underlying population, selection bias related to discrepancies in documentation compliance and adverse event awareness, as well as differences in stress examinations and contrast agent usage. In this context, it has to be highlighted that the most frequent severe AAEs in our study were arrhythmias, which might be attributable to underlying cardiac disease in this specific patient population as well.

Our study population of 72,839 patients is the largest cohort evaluating AAEs in CMR imaging when compared to 37,788 [5] or 17,767 [13] patients in the EuroCMR Registry. Results on a lower AAE rate in the EuroCMR Registry might be related to a smaller cohort with less contribution—mainly academic centers and countries (57 versus 263 centers and 15 versus 32 countries). Further, our study evaluated a significantly larger proportion of male subjects (65% versus 62.2% male) [6].

In our study, the majority of patients demonstrated pathological CMR scans. While comparable data is unavailable from the EuroCMR Registry, one may speculate that the number of pathologic CMRs was lower than that in our cohort.

To the best of our knowledge, our study is the first to explicitly evaluate the interaction and occurrence of AAEs in combined stress CMR examinations with usage of GBCA. Only one EuroCMR Registry study provided information on stress CMR imaging, with severe AAEs (0.1%) reported exclusively in the group of patients receiving stress CMR imaging [22]. Still, EuroCMR Registry studies did not provide information on the exact pharmacological stressor used. In the ESCR MRCT Registry, regadenoson and dobutamine account for almost 10% of stress imaging studies, which are associated with more AAEs, i.e., due to a prolonged half-life and different modes of action as compared to adenosine [23].

There were considerable differences in the GBCA distribution of our study (Fig. 1) compared to the most recent EuroCMR Registry AAE study [6] with administration of gadopentetate (33.8%), gadobutrol (24.8%), gadodiamide (16.2%), and gadoteric acid (11.2%). However, the distribution of GBCA among vendors is substantially influenced by submitting centers and does not necessarily reflect the actual market shares in Europe.

Finally, discrepancies in documentation compliance and adverse event awareness might contribute to the higher number of AAEs observed in the radiological ESCR MRCT Registry as compared to the cardiology-initiated EuroCMR Registry: cardiologists might underrate or fail to report moderate AAEs such as dyspnea in the setting of stress CMR imaging.

After multivariable statistical adjustment, pharmacological stressors and GBCA subtypes emerged as independent, significant predictors of acute adverse events. Patients subjected to stress imaging were more likely to develop acute adverse events than those without stress imaging. Across different pharmacological stressors, regadenoson and dobutamine showed an increased likelihood when compared to adenosine. In line with the recent literature, adverse event rates were higher among regadenoson and limited to mild and moderate events, mainly dyspnea [24, 25]. This difference could be attributed to the pharmacokinetic properties of regadenoson with a comparably long, tri-phasic half-life time [23]. However, increased respiratory adverse events could originate from selective regadenoson administration in patients with pulmonary diseases as well [26]. Since pulmonary diseases were not evaluated in the registry, residual confounding might have distorted our results.

Our findings on higher adverse event rates disagree with results from the EuroCMR Registry, showing lower adverse event rates in stress imaging [5, 13]. However, the authors did not statistically adjust for potential confounders such as age, gender, and imaging indications.

Although not the main aim of our study, gadoteridol had a higher likelihood than gadobutrol to be associated with acute adverse events. Disparities in GBCA molecular structure and chelate stability were suggested as potential explanations, supported by a recent meta-analysis showing increased AAE risk for linear ionic GBCA [3, 14, 27, 28]. The missing statistical influence of GBCA molecular properties in our study might be attributable to the overall low number of AAEs and consecutively low statistical power in certain subgroups. Still, comparisons between the GBCA subtypes were adequately powered to detect very small effect sizes at 90–95% power with an alpha level of 0.05, even considering the smallest GBCA strata [29]. Concerns that imbalances in GBCA strata sizes, ranging from approximately 2250 to 40,000 patients, biased our results are not supported by the current statistical literature [30].

Interestingly, GBCA-associated adverse events varied across pharmacological stressors. For example, gadobutrol-associated adverse event rates ranged from 0.52% for adenosine to 4.56% for regadenoson, while event rates were balanced for gadoteric acid (0.3% versus 0.47%). This might be explained by a potentiation of contrast media–associated acute events, depending on different pharmacological stressors, but has not been described in the literature so far.

Although not statistically significant, subgroup analyses suggest that higher GBCA volume might contribute to our findings. For example, higher GBCA volumes were evident in the regadenoson subgroup with high AAE incidence. Further, severe AAEs were mainly categorized as physiologic and received larger GBCA volumes compared to other strata. Especially considering our patient cohort with known or suspected cardiac impairment, intravenous fluid administration might intuitively correlate with AAE incidence, as described for patients with acute decompensated heart failure [31]. We aim to prospectively collect additional data on the total amount of intravenous fluid administration with updated versions of the MRCT Registry to further elucidate these pathomechanisms.

In contrast to earlier studies, we analyzed routinely used GBCAs and statistically adjusted for confounders. Differences in the underlying patient population or varying AAE ascertainment methods across the submitting centers were considered in statistical analyses. Reporting on AAEs in the context of CMR imaging was standardized in the ESCR MRCT Registry and completed by physicians. Finally, our results proved robust upon sensitivity analyses.

Still, our study is not devoid of limitations. Information on GBCA concentration and volume were non-mandatory and missing in the majority of cases, which might have biased results on GBCA-specific AAE profiles. There was no information on long-term follow-up, limiting the validity for late-onset adverse events such as nephrogenic systemic fibrosis [32]. Despite a large overall sample size, single stressor and GBCA subgroups might be undersampled to promote generalizable conclusions. The MRCT Registry was not specifically designed to evaluate GBCA safety, and selection biases must be considered regarding the non-random fashion in which stressor and contrast agents were administered. For example, cardiac imaging for cardiomyopathy was conducted mainly without pharmacological stressors, whereas stress imaging was performed in the majority of patients with suspected or known CAD. Furthermore, patients receiving regadenoson might have been selected due to underlying pulmonary diseases, thus explaining a higher rate of respiratory acute adverse events in the regadenoson subgroup. The proposed interactions between stressor and GBCA need therefore to be validated in randomized studies, controlling for underlying clinical conditions. Moreover, results on GBCA-specific AAE rates might be biased by selective data collection and reporting due to the registry design and the retrospective analysis of our study.

Finally, no information was obtained regarding the patients’ renal function and previous history of adverse reactions to contrast media, which may have biased our results.

Our results suggest that gadolinium-based contrast agents are generally safe for application in cardiac MR imaging. Observed acute adverse event rates are within the range reported for general radiology and differ across GBCA, stress test imaging, and pharmacological stressor, as well as main indications. Higher volume intravenous fluid administration in patients with cardiac impairment might at least partially contribute to our findings. Further, pharmacological stressors might potentiate GBCA-associated adverse events. To the best of our knowledge, we provide the largest and most comprehensive study for acute adverse events in GBCA-enhanced cardiac MR imaging, evaluating a variety of routinely used GBCA in a multinational, multicenter setting.