The deposition of gadolinium (Gd) in the brain following Gd-based contrast agent (GBCA) administration has resulted in strict labeling changes for all GBCAs marketed in the United States (U.S. Food and Drug Administration, 2017). Despite attention in the literature since GBCAs were first associated with hyperintensity in the dentate nucleus (Kanda et al., 2014), the mechanism of deposition is still not well understood (McDonald et al., 2018). A leading proposed mechanism involves uptake into the cerebral spinal fluid (CSF) through a glymphatic pathway, which explains the entry of highly polar intact GBCA complexes into the brain by bypassing of the lipoidal blood-brain barrier (Deike-Hofmann et al., 2018); the assumption of this mechanism is that some intact GBCA will simply remain in the brain, slowly releasing Gd, which will redistribute to metal-storage sites in the dentate nuclei, globus pallidus, and thalamus. The clinical significance and consequence of retained Gd in the brain is unknown (Gulani et al., 2017). Currently, there are no studies that have demonstrated a causal link between Gd deposition in the brain and clinical symptoms. However, patient reporting of neurological symptoms post-GBCA administration include headaches, vision changes, and hearing changes (Olchowy et al., 2017). Furthermore, there is evidence of Gd encephalopathy secondary to intrathecal administration and toxicity (Kapoor et al., 2010). The glymphatic pathway mechanism is well-supported by analyses of CSF Gd in both rats and humans (Nehra et al., 2018; Taoka et al., 2018). Support for the CSF-based mechanism has contributed to reduced attention on earlier proposed mechanisms related to iron (Fe) and transferrin (Tf) transport (Prybylski et al., 2016), which has so far only been suggested by the strong correlation of Gd deposition sites and Fe storage sites in the brain (Rasschaert et al., 2018). The possibility that both candidate mechanisms may play a significant role in Gd deposition in the brain has not been explored and could have major implications for clinical risk factors of Gd deposition in the brain.

Fe homeostasis is a complex process involving a myriad of transporters and a family of transport proteins (Duck and Connor, 2016). In plasma, Fe is 98%–100% bound to transferrin, but more than 50% of Fe-binding sites on transferrin are unoccupied. Non-transferrin bound Fe is rapidly taken into peripheral tissues where it can cause oxidative damage (Craven et al., 1987); maintaining low transferrin saturation prevents this outcome and facilitates proper packaging of ferric iron (Fe³⁺) into the storage protein ferritin in tissues. In tissues, Fe turnover is a result of using Fe stores as enzymatic cofactors, but in the brain, Fe is stored in abundance so there is very slow clearance of Fe (Dallman and Spirito, 1977). Gd and ferric Fe share the same charge (3+), but Gd has twice the ionic radius and a higher coordination number. These differences likely explain why Gd only binds a single Fe-binding site on human transferrin and with relatively low affinity (logK_Tf:Gd = 6.8, logK_Tf:Fe(III) = 21) (Zak and Aisen, 1988). However, there is still a strong correlation between deposition sites for Gd and Fe in the brain in humans and rats (Prybylski et al., 2016; Rasschaert et al., 2018), so it is unlikely that Gd deposition is entirely unrelated to Fe transport and storage mechanisms.

Despite evidence suggesting the importance of Fe in Gd deposition, no studies have been reported to show anything but a correlative relationship. Gd deposition within the dentate nuclei, globus pallidus, and thalamus following multiple GBCA exposures have been confirmed. These iron-rich brain structures are associated with neurodegenerative disorders and manganese accumulation (Swaminathan, 2016). Further, the Fe-Gd correlation is specific and not just a general metal storage error, as no such correlation exists for Cu-Gd or Zn-Gd (Rasschaert et al., 2018). In this study, we sought to determine if there is a causative relationship between iron status and Gd deposition/distribution by using modified iron diets in animals in order to model transient iron deficiency and iron overload. We also examined the impact of time on any Fe-dependent Gd effects.

The tissue concentrations of Fe (excluding brain) in both washout groups were significantly correlated with level of Fe in the diet (Figure 3), showing that Fe status had been affected by the intervention. There was no difference by between Fe concentrations within diet level when comparing washout groups, except in spleen where the concentration of iron was slightly lower in the 3-week washout of the high iron group (mean ± standard deviation 2049 ± 948 versus 2,670 ± 598 μg/g in high iron 3-week versus 3-day). In the brain, there was no impact of dietary Fe on total brain iron. For all tissues in both washout groups (3-day and 3-week), there was no significant correlation between Gd content and Fe in the diet (Figure 4); the closest to achieving significance was the brain Gd concentration in the 3-week washout group (F = 2.85, p = 0.11). In most tissues, the 3-week washout was associated with clearance of Gd relative to the 3-day washout group, but there was significant Gd accumulation in the brain, particularly in the -Fe group (Figure 5).

The distribution of Gd in the brain was significantly impacted by Fe levels in the diet for the 3-week washout group, but there was no diet-level effect in the 3-day group (Figure 6). Specifically, ∼4-fold higher Gd concentrations were observed in the olfactory bulb for the -Fe group compared to the FeC and +Fe groups (adjusted p < 0.0001 for both). The difference in distribution of Gd was significant between the low Fe diet and the normal/high Fe diets. Non-significant trends were observed in other slices, including B (includes a portion of the olfactory bulb) and F (includes the deep cerebellar nuclei), all with more Gd in the -Fe group. In the 3-day group, there was too much blood contamination of the brain to determine the correlation between Fe and Gd concentrations in the brain slices, resulting in a flat distribution of iron across slices. In the 3-week study, the Fe and Gd concentrations in each slice were adequately correlated (r² > 0.6 and p < 0.05 for all Fe groups; Figure 7).

Mean tissue Fe concentrations in the 3-week washout group correlated well with mean tissue Gd concentrations, excluding in the whole brain, stratified by diet group and accounting for variability. These correlations were not significant for any tissue in the 3-day washout group. The iron-Gd correlation was strongest in the liver, where the r² was 1.000 with a negative slope, and poorest in the total brain concentration, where the r² was 0.054; kidneys and spleen had r² values of ∼0.6 with negative slopes and femur had an r² of 0.892 with a positive slope. Data for Fe-Gd correlation across tissues is not shown. Groupwise correlations of means (considering variability) were also observed across tissue concentrations of Gd in the 3-week group stratified by iron diet (Table 1). Gd concentrations in the femur had a strong, negative correlation with Gd concentrations in other tissues, though no correlation was significant when adjusted for multiple comparisons.

The study demonstrated a significant effect of dietary Fe on the long-term retention and distribution of Gd in tissues, particularly the olfactory bulb. GBCAs have demonstrated deposition of Gd in brain regions that are critical for advanced cognition and executive function, behavior and affect, and sensorimotor coordination (Hua et al., 2022). The lack of a diet-level effect on short-term deposition of Gd suggests that at that stage there is no competition with Fe for transferrin, further contributing to evidence that GBCAs deposit in tissues as intact chelate (Robert et al., 2018). However, the long-term results challenge the hypothesis that prolonged Gd retention in a given tissue is caused by release of Gd from GBCAs that had already deposited in that tissue; i.e., we have demonstrated that released Gd re-enters the circulation, as moving from one tissue to another requires re-entering circulation (unless the tissues are in physical contact), and is re-distributed as free Gd ions would be. This is significant in that it suggests a mechanism for the strong correlation between Fe and Gd concentrations in discrete brain structures for less stable linear GBCAs, while the correlation is non-existent for more stable macrocyclic agents (Rasschaert et al., 2018): deposition of the latter is non-specific while current evidence suggests linear agent deposition is primarily by mobilization of peripheral Gd by Fe transport mechanisms. This is further supported by the amount of Gd deposited from linear GBCAs which is bound in large, macromolecular complexes (Gianolio et al., 2017), which may be further non-specific selection of Gd³⁺ for Fe³⁺. However, additional questions are raised about the intracellular distribution of Gd; the divalent metal transport 1 (DMT1) protein is essential for the uptake of transferrin-bound Fe to the brain and it has no cross-reactivity with trivalent ions (Illing et al., 2012), so how would Gd enter the cytoplasm from an endocytic vesicle? The reduction potential for Gd³⁺/Gd²⁺ is much lower than that of Fe³⁺/Fe²⁺, so it is very unlikely that Gd can undergo the same reduction necessary to transport Fe by DMT1. The intracellular distribution of Gd is an important open question about Gd deposition, and one that could be answered with antibody-based methods that have been developed for Fe speciation (Moos and Morgan, 1998).

The olfactory bulb accumulation of Gd observed in this study has not been observed in similar term studies. The accumulation was only significant for the -Fe group, but there was an appearance of higher Gd in the other groups (Figure 5). This is inconsistent with a similar, kinetic study in which rats were given five weekly, high-dose gadodiamide loading sequence, and there was a significant ∼20% reduction in brain Gd in rats euthanized a month after the last dose compared to rats euthanized a week after the last dose (Robert et al., 2018). To be consistent with the present study, the Gd would be expected to slightly increase in the 1-month group. The difference could be attributable to the time lag from the last dose (7 days vs. 3 days), weekly rather than daily GBCA dosing, or the different gadodiamide formulations (Omniscan^® vs. gadodiamide without excess ligand). A combination of the time lag and weekly dosing are most likely to have caused the discrepancy as some doses had well over a month to equilibrate in the body, meaning any redistribution of peripheral Gd to the brain (as seen in the Fe study) would have occurred before the first group of rats was euthanized. It is also possible that Omniscan^® with its increased stability resulting from excess ligand (Tweedle et al., 1995) has greater early, glymphatic-mediated brain deposition, and the early release of Gd from unformulated Gd results in more peripheral Gd deposition.

The correlations between tissue Fe concentration and tissue Gd concentrations, or between tissue Gd concentrations, further support the conclusion that there is a relationship between Fe and Gd storage. The negative correlation between Gd in femur vs. other tissues suggests that the mechanism behind long-term Gd deposition involves an equilibrium of Gd incorporated into bone and released from bone, similar to calcium; with increased Gd-binding sites on the transferrin in plasma (as Fe levels go from high to low), that equilibrium shifts to more release of Gd from bone, resulting in transferrin-mediated Gd deposition in other tissues. Equilibria may exist in other tissues but since negative correlations are only noted in the femur, the bone turnover equilibrium must be more pronounced than in other tissues. This may have implications for conditions associated with increased bone turnover and patients with risk factors for those conditions. It has already been shown that Gd concentrations are decreased in osteoporotic patients relative to other patients exposed to GBCAs (Darrah et al., 2009); is it possible these patients would have higher amounts of Gd in their liver and brain? These findings also have implications for patients with comorbid Fe deficiency and osteoporosis, or other conditions of increased bone turnover. Since Fe deficiency can induce bone turnover (Toxqui and Vaquero, 2015), there may be a considerable number of patients at risk of excess non-bone Gd deposition.

This study had several limitations, most significantly a lack of additional contrast agents under investigation. If a stable, macrocyclic agent had been used and the same trend was observed with respect to dietary iron, then transferrin-related changes may not impact Gd deposition. It is known, however, that Gd from gadodiamide binds macromolecules in vivo, whereas most evidence suggests macrocyclic agents only deposit intact and never release Gd (Rasschaert et al., 2018; Robert et al., 2018). Additionally, several trends identified in the correlation analyses were strong, but not significant in that comparison or in related Fe diet comparisons. This suggests that the study was underpowered to detect those differences, though it was clearly powered sufficiently to detect differences in brain distribution, which was a primary outcome under investigation. Finally, the relatively short washout of 3 weeks for the long-term cohort may not be enough to extrapolate how transient changes in Fe status impact Gd deposition, however a 3 week washout is on par with washout periods used in other animal studies intended to draw comparisons to clinical observations of long-term retention (Davies et al., 2021). This limitation was considered negligible since in the brain kinetic study referenced earlier (Robert et al., 2018), Gd concentrations in the brain had essentially reached plateau levels 1 month after the last dose of gadodiamide. The addition of male rats would have been valuable to the study considering sex differences in Fe handling and bone turnover.

Fe deficiency increases the concentration of Gd in Fe storage sites of the rat brain, primarily the olfactory bulb, implying that some Gd deposition is the result of Fe transport mechanisms, likely transferrin. The clinical relevance of these findings may extend to patients at risk of Fe deficiency (e.g., in renal failure) and increased bone turnover. Rare earth metal retention in the olfactory neurons and iron homeostasis disruptions via MRI contrast agents could lead to anosmia, which is important for diagnosticians to consider.