Although MRI is considered a safe, non-invasive diagnostic imaging technique, in contrast to approaches based on ionising radiation exposure, several concerns about potential MRI-related health risks have arisen in the past few years and have not yet been sufficiently clarified to date. As a potential genotoxic effect of MRI would have a large impact on the clinical practice of MRI, several studies have been conducted in the past decade. Whereas some studies did not find any significant increase in genetic damage after MRI exposure, other studies reported a significant genotoxic MRI-related effect [11,12,13,14,15,16,17,18,19, 24, 28]. These studies, however, show a large diversity regarding field strengths, exposure parameters, and types of genotoxicity analysis. In addition, no established hypothesis for a potential mechanism can explain the results of the studies where significant effects were found.
Furthermore, it is known that no single genotoxic endpoint (e.g., single/double-stand breaks, chromosomal aberrations, micronuclei) allows one to determine the carcinogenic risk of an agent. Consequently, further studies that try to investigate possible underlying mechanisms of genotoxicity should divide one sample into multiple aliquots in order to investigate several end-points of genotoxicity and cytotoxicity at the same time . In this context, researchers also have to look for other possible genotoxic endpoints, such as oxidative stress, as has been reported by Erdamar et al. in 2014 .
As two studies that observed DNA damage were conducted after contrast-enhanced MRI [13, 24], a potential cause could also lie in a possible interaction mechanism with contrast media or in an interaction of contrast media with MRI. Therefore, in this study we examined the genotoxic and cytotoxic impact of different GBCAs on human blood lymphocytes and analysed a potential additive or synergistic effect when simultaneously exposed to 7-T MRI.
At our institution, ultra-high-field MRI is applied for research purposes only and administration of contrast agents is prohibited in studies conducted in vivo on humans. Thus, the current investigations had to be performed under in vitro conditions. However, the use of isolated, unshielded lymphocytes, the constant presence of high GBCA concentrations and the application of 7-T MRI combined with the maximum permissible switched gradient and specific absorption rate all present enhanced exposure conditions in these samples compared to in vivo examinations, and would presumably also maximised the level of cell damage and detectability.
Genotoxicity was determined by γH2AX staining, which is one of the most sensitive biomarkers for DNA DSB assessment. Our experiments confirmed the induction of γH2AX foci after CT exposure and an additional dose-dependent increase in the presence of ICA, while ICA alone did not lead to γH2AX formation. As has also been shown in previous publications [20,21,22], the extent of DNA damage was dependent on irradiation dose as well as on ICA concentration. DSB induced by unenhanced CT were mainly repaired and did not alter apoptosis or proliferation rate, whereas ICA-induced cytotoxicity was partially enhanced for high dose ICA in combination with CT. In contrast, GBCA treatment alone and simultaneous 7-T MRI exposure showed no evidence of enhanced γH2AX levels in peripheral lymphocytes. While Fiechter et al.  reported a significant increase in γH2AX after gadobutrol-enhanced MRI exposure, our current results support the findings of Brandt et al.  and Reddig et al. , who reported no induction of γH2AX foci in patients 5–30 min after contrast-enhanced MRI.
As Gd- and GBCA-induced cytotoxicity has been described for various cell types [5, 26, 30], we determined the level of cell death induced by different structural classes of GBCAs and investigated a potentially enhancing effect in combination with 7-T MRI. The apoptosis rate of lymphocytes was analysed 24 h after MRI exposure while GBCAs were still present. At a dose of 2 mM, a significant increase in cell death was only observed for the investigated linear, non-ionic gadodiamide, while a concentration of 20 mM induced apoptosis for all three linear GBCAs. This cytotoxic effect occurred independently of MRI exposure and was detectable to a stronger extent by the level of DNA synthesis determined after 72 h in activated lymphocytes. At a dose of 20 mM, analysed classes of linear GBCAs led to a strong reduction in the proliferation rate, whereas no decrease was observed for the two macrocyclic GBCAs included in our study. However, additional markers need to be assessed to clarify this observation in activated lymphocytes.
A surprising finding was a slight decrease in apoptosis when lymphocytes were cultured in the presence of the macrocyclic gadobutrol (Fig. 2b). This decrease seems difficult to explain, as no mechanism for potential MRI-induced primary or secondary genotoxic or cytotoxic effects is known. To investigate this effect more deeply in further studies, it could be helpful to perform additional apoptosis assays, such as caspase 3 activity, which was not part of this study.
Furthermore, the exposure conditions required for cytotoxicity assessment in our in vitro experiments differed in comparision to GBCA pharmacokinetics in vivo. Upon injection in vivo GBCA rapidly distributes to the extracellular space and is eliminated from the blood with a half-life of approximately 1.5 h in subjects with normal renal function [31, 32]. Therefore, the drawing of reliable conclusions from our in vitro data for clinical routines is difficult.
As shown by Cho et al. , exposure of human lymphocytes to non-chelated GdCl3 led to a dose-dependent induction of apoptosis and DNA damage, which were further enhanced when samples were co-exposed to Gd and an extremely low-frequency electromagnetic field. Our cytotoxicity data compare well with previously described dissociation characteristics of toxic free Gd3+ from chemically different GBCA chelates. In general, macrocyclic GBCAs are described as being more stable than linear compounds [5, 33]. Analysing the Gd3+ release of multiple GBCAs in human serum over a period of 15 days, Frenzel et al.  observed an increase of free Gd3+ from ionic and especially non-ionic linear GBCAs, whereas all macrocyclic GBCAs remained stable. A correlation between osmolality of GBCA and cytotoxicity, as described for porcine renal epithelial cells by Heinrich et al. , was not observed in our study. Our results confirmed increased toxicity induced by the group of investigated linear GBCAs compared to the two analysed macrocyclic ones. Nevertheless, further in vitro and in vivo studies are needed to evaluate all commercially available MRI contrast agents and to detect possible mechanisms of toxicity or potential interaction in combination with high- and ultra-high-field MRI.
Limitations of this study include the circumstances that only a limited amount of analyses could be conducted within this experimental setup, and no conclusions regarding different cell types, different DNA lesions or long-term effects in vivo can be drawn. Furthermore, effects due to ingredients other than the pure substance of the GBCA cannot be fully excluded. Since no exact mechanisms about potential MRI-induced primary or secondary genotoxic or cytotoxic effects are known, it is difficult to choose the most sensitive assay and best point in time. Until now, experimental procedures are mainly motivated by studies investigating ionising radiation. The exposure conditions required for cytotoxicity assessment in our in vitro experiments differed regarding GBCA pharmacokinetics in vivo, as mentioned above. Therefore, the genotoxic and cytotoxic impact of different GBCAs in MRI needs to be further examined, in particular regarding Gd tissue retention and possible pathological effects, especially under more reliable in vivo conditions.
In conclusion, our in vitro data demonstrate no induction of γH2AX foci in isolated human lymphocytes after contrast-enhanced 7-T MRI exposure with respect to the five investigated GBCAs. Enhanced cytotoxicity was observed for the three tested linear GBCAs compared to the two macrocyclic ones, but no further amplified cell damage was determined when GBCA exposure was combined with 7-T MRI.