Comprehensive Analysis of the Spatial Distribution of Gadolinium, Iron, Manganese, and Phosphorus in the Brain of Healthy Rats After High-Dose Administrations of Gadodiamide and Gadobutrol

Abstract

Objectives

After the administration of gadolinium-based contrast agents (GBCAs), residual gadolinium (Gd) has been detected in a few distinct morphological structures of the central nervous system (CNS). However, a systematic, comprehensive, and quantitative analysis of the spatial Gd distribution in the entire brain is not yet available. The first aim of this study is to provide this analysis in healthy rats after administration of high GBCA doses. The second aim is to assess the spatial distributions and possible Gd colocalizations of endogenous iron (Fe), manganese (Mn), and phosphorus (P). In addition, the presence of Gd in proximity to blood vessels was assessed by immunohistochemistry.

Materials and Methods

Male rats were randomly assigned to 3 groups (n = 3/group): saline (control), gadodiamide (linear GBCA), and gadobutrol (macrocyclic GBCA) with cumulative Gd doses of 14.4 mmol/kg of body mass. Five weeks after the last administration, the brains were collected and cryosectioned. The spatial distributions of Gd, Fe, Mn, and P were analyzed in a total of 130 sections, each covering the brain in 1 of the 3 perpendicular anatomical orientations, using laser ablation coupled with inductively coupled plasma mass spectrometry. Quantitative spatial element maps were generated, and the concentrations of Gd, Fe, and Mn were measured in 31 regions of interest covering various distinct CNS structures. Correlation analyses were performed to test for possible colocalization of Gd, Fe, and Mn. The spatial proximity of Gd and blood vessels was studied using metal-tagged antibodies against von Willebrand factor with laser ablation coupled with inductively coupled plasma mass spectrometry.

Results

After administration of linear gadodiamide, high Gd concentrations were measured in many distinct structures of the gray matter. This involved structures previously reported to retain Gd after linear GBCA, such as the deep cerebellar nuclei or the globus pallidus, but also structures that had not been reported so far including the dorsal subiculum, the retrosplenial cortex, the superior olivary complex, and the inferior colliculus. The analysis in all 3 orientations allowed the localization of Gd in specific subregions and layers of certain structures, such as the hippocampus and the primary somatosensory cortex. After macrocyclic gadobutrol, the Gd tissue concentration was significantly lower than after gadodiamide. Correlation analyses of region of interest concentrations of Gd, Fe, and Mn revealed no significant colocalization of Gd with endogenous Fe or Mn in rats exposed to either GBCA. Immunohistochemistry revealed a colocalization of Gd traces with vascular endothelium in the deep cerebellar nuclei after gadobutrol, whereas the majority of Gd was found outside the vasculature after gadodiamide.

Conclusions

In rats exposed to gadodiamide but not in rats exposed to gadobutrol, high Gd concentrations were measured in various distinct CNS structures, and structures not previously reported were identified to contain Gd, including specific subregions and layers with different cytoarchitecture and function. Knowledge of these distinct spatial patterns may pave the way for tailored functional neurological testing. Signs for the localization of the remaining Gd in the vascular endothelium were prominent for gadobutrol but not gadodiamide. The results also indicate that local transmetalation with endogenous Fe or Mn is unlikely to explain the spatial patterns of Gd deposition in the brain, which argues against a general role of these metals in local transmetalation and release of Gd ions in the CNS.

In 2014, Kanda and colleagues¹ were the first to report abnormally increased signal intensity in the dentate nucleus and globus pallidus (GP) on unenhanced T1-weighted brain magnetic resonance images (MRIs) in patients with normal kidney function who had previously received repeated injections of the linear gadolinium-based contrast agents (GBCAs) gadodiamide or gadopentetate dimeglumine. The vast majority of studies corroborated the finding of T1 hyperintensities after intravenous administration of linear GBCAs, but not after macrocyclic GBCAs in clinical cohorts with different age, sex, medical (pre)conditions, and GBCA dosage regimen.^2–7

At the same time, postmortem analyses by inductively coupled plasma mass spectroscopy (ICP-MS) of dissected dentate nucleus and GP from patients who had received multiple injections of linear GBCA revealed that the T1 hyperintensities reflect tissue depositions of gadolinium (Gd).^3,6 Subsequent studies found T1 hyperintensities also in other CNS structures including parts of the thalamus, the superior colliculus, the putamen, and the substantia nigra, whereas no hyperintensities were found in the white matter.^8–10

Comparison of T1-weighted images of patients with postmortem ICP-MS data indicated that the highest Gd concentrations were found in the dentate nucleus and the GP, that is, the structures with the greatest T1 hyperintensities, but that low levels of Gd are more widespread in the brain than are detectable by MRI. For example, ICP-MS found Gd in structures such as the frontal lobe cortex and the pons; trace amounts were even detected in white matter.^3,6

The T1 hyperintensities were consistently much more pronounced and the Gd concentrations in postmortem specimen much higher after linear versus macrocyclic GBCAs, which is attributed to the lower kinetic inertness of the former.^11,12 Based on these findings, in 2017, the European Commission suspended marketing authorization of linear extracellular multipurpose GBCAs and restricted the use of gadobenic acid to liver scans, and the US Food and Drug Administration requested vendors to change the product safety labeling of their GBCAs. At the same time, the US Food and Drug Administration encouraged to undertake both preclinical and clinical investigations into the phenomenon of CNS Gd uptake, and to determine whether or not the presence of Gd in the brain may adversely affect its functions.^7,12,13

Although studies consistently found Gd presence of varying extent in many organs after GBCA administration,^14,15 a causal relationship to specific clinical symptoms has hitherto only been established for nephrogenic systemic fibrosis in renally impaired patients.^16,17 No causal relationship, however, has been established so far between the presence of Gd in the CNS and potential clinical adverse effects. Yet, starting in 2016, Semelka et al^18,19 postulated that a variety of neurologic symptoms self-reported by patients after exposure to GBCA would be caused by Gd accumulation and proposed a new disease entity termed Gd deposition disease. In 2021, the American College of Radiology Committee on Drugs and Contrast Media proposed to use the term symptoms associated with Gd exposure instead of Gd deposition disease,²⁰ indicating that these symptoms are not unambiguously caused by GBCAs.

Several preclinical rat studies used laser ablation ICP-MS (LA-ICP-MS) to localize Gd in various preselected structures of the gray matter.^21–25 This method allows to generate high-resolution Gd maps of tissue slices irrespective of the molecular species (chemical form) of retained Gd. The identification of CNS structures that accumulate Gd is important, because tailored neurological tests can then be used that address the specific functions of these structures in order to ascertain whether Gd presence impacts these functions. The first purpose of the present study was to provide a systematic, comprehensive, and quantitative LA-ICP-MS analysis of the spatial Gd distribution of the entire brain in all 3 perpendicular anatomical orientations (sagittal, horizontal, coronal) after high cumulative doses of the linear GBCA gadodiamide and the macrocyclic GBCA gadobutrol.

To date, it is not fully understood in which chemical species Gd is retained in the CNS and why it is present in several distinct structures of the gray matter but not in others. One possible pathway is the release of the Gd³⁺ ion from its linear ligand and uptake by some endogenous ligands.²⁶ This process of transmetalation could occur in tissues with high levels of endogenous metals.^12,23,27 The deep cerebellar nuclei (DCN), GP, and striatum, that is, the first structures in which residual Gd was detected, are known to be rich in iron (Fe).^28–30 By means of juxtaposed or overlaid element maps obtained by LA-ICP-MS, colocalizations of Gd and Fe have been observed.^21,23–25 Thus, transmetalation has been proposed as a major reason behind the spatial patterns of residual Gd.^12,23,27 The second purpose of the present study was to test this hypothesis by image overlays and quantitative analysis of the whole-brain spatial distributions of Gd, Fe, and Mn.

The cellular environment of the residual Gd in certain brain structures is still poorly understood. The third aim of the present study was the localization of Gd with regard to the vasculature by investigating the spatial Gd distribution together with an endothelial marker.

MATERIALS AND METHODS

Animals, Treatment, and Tissue Preparation

A total of 9 male Han-Wistar rats (Crl:WI; body mass 230–250 g) were obtained from Charles River Laboratories (Sulzfeld, Germany). All investigations were approved by the local Animal Welfare authority in accordance with the German Animal Protection Law, and the experiments were carried out in accordance with the approved guidelines. The animals were kept under standard laboratory conditions, and food and water were provided ad libitum.

Rats were randomly assigned to 3 treatment groups (n = 3 per group) and received either gadodiamide (linear GBCA, Omniscan; GE Healthcare Buchler & Co, Braunschweig, Germany), gadobutrol (macrocyclic GBCA, Gadovist; Bayer Vital, Leverkusen, Germany), or isotonic saline solution (used as control; B. Braun Melsungen AG, Melsungen, Germany). Animals received 8 intravenous injections, each at a dose of 1.8 mmol Gd per kilogram of body mass (equivalent to a triple standard dose in humans normalized according to body surface area)³¹ for 4 consecutive days per week during a 2-week period, resulting in a cumulative dose of 14.4 mmol Gd per kilogram of body mass. Five weeks after the last injection, the rats were killed by exsanguination under anesthesia with 1.5% isoflurane (Baxter, Unterschleissheim, Germany).

Whole rat brains including the olfactory bulb (OB) were harvested and were divided into the hemispheres along the medial sagittal plane using a disposable blade. The resulting hemispheres were embedded in Tissue-Tek OCT compound (Sakura Finetek, Alphen aan den Rijn, the Netherlands) and snap frozen in isopentane with dry ice. The hemispheres were subjected to cryosectioning (CM1950; Leica Biosystems GmbH, Nussloch, Germany) along the 3 anatomical orientations (sagittal, horizontal, and coronal; 1 hemisphere per orientation and group). In each orientation, multiple tissue slices of 10 μm thickness (with an average spacing of approximately 1.5 mm) were collected on glass slides (Menzel Glaeser, Braunschweig, Germany), dried at 37°C, and subsequently stored at −20°C. For each group and in each orientation, between 12 and 15 tissue slices were analyzed by LA-ICP-MS.

Laser Ablation Inductively Coupled Plasma Mass Spectroscopy

The spatial distribution of the elements Gd, Fe, Mn, and P in the brain slices was determined by a laser ablation system (NWR 213; New Wave Research, Fremont, CA) coupled to an ICP-MS (iCAP RQ ICP-MS; Thermo Fisher, Bremen, Germany).

Laser ablation was performed as previously described in detail.^21,32 Briefly, continuous-line ablation mode was used, and the laser energy was optimized daily to attain approximately 2 J/cm², which leads to complete local ablation of the sample material. The ablation spot size was 40 × 50 μm² with no distance between lines. The laser frequency was 20 Hz, and the scan speed was 100 μm/s, which resulted in a pixel dimension of 11.5 × 50 μm² and a voxel dimension of 11.5 × 50 × 10 μm³. The ablated tissue material was transferred to the ICP-MS by a constant flow of helium (900 mL/min), and the isotopes ¹⁵⁸Gd, ⁵⁷Fe, ⁵⁵Mn, and ³¹P were measured quasi-simultaneously. To quantify the tissue metal concentrations, gelatin standards with 4 defined concentrations of the respective elements (Gd: 0.0, 1.1, 21.6, and 54.4 nmol/g; Fe: 14.7, 22.0, 228.9, and 562.7 nmol/g; Mn: 1.0, 1.8, 22.7, and 55.9 nmol/g) were used as previously described.³² The limit of detection for Gd was ≤0.2 nmol/g wet tissue. The analysis of a single tissue slice took from 3 to 18 hours, depending on its size.

Image Generation, Quantification of Element Concentrations, and Assignment of Regions of Interest

Images visualizing the spatial distribution of the analyzed elements within the brain were generated using the software ImaJar, generously provided by R. Schmid and U. Karst (University of Münster, Münster, Germany). The raw data from all ablation lines from the ICP-MS were converted to a pixel graphic in which each pixel represents the respective ablation spot with its signal intensities corresponding to the tissue concentration of the respective elements. These are displayed in color-coded scales. In images with very low or absent Gd signals, the tissue borders based on P images were transferred to the Gd images. To estimate the amount of residual Gd in the whole brain, the average concentration in an entire medial sagittal tissue slice comprising the whole brain tissue including the white matter was measured. For identification of individual brain structures, images were meticulously aligned to the 3-dimensional rat brain atlas by Paxinos and Watson,³³ which delineates brain structures based on their different cytoarchitecture and function. This allowed freehand placement of regions of interest (ROIs) covering distinct brain structures using ImaJar. Regions of interest were labeled in accordance with the rat brain atlas.³³ Across all groups and elements, ROIs were placed primarily according to Gd hotspots observed after gadodiamide treatment. The anatomical orientation of the tissue slices for ROI placement was chosen based on the best possible representation of the respective brain structure such that most ROIs were placed in the largest possible cross-section of each respective brain structure. In addition, where possible, we placed ROIs for multiple different structures in a given tissue slice, which was most feasible in sagittal slices. Average concentrations of Gd, Fe, and Mn were calculated for each ROI. Overlay images of two elements were created using green and red monochrome scales. Colocalizations of both elements yield yellow pixels.

Immunohistochemistry LA-ICP-MS

To evaluate the validity of the approach, we compared the tissue distribution of target proteins detected by metal-labeled antibodies and LA-ICP-MS with the distribution obtained by using standard immunofluorescence (IF) analysis with fluorochrome-conjugated antibodies on serial brain sections (Supplemental Digital Content, Fig. S1, http://links.lww.com/RLI/A887). Staining of frozen brain sections was performed as described,³³ modified for indirect detection using secondary metal- or fluorochrome-tagged antibodies. Briefly, sections were fixed in cold acetone (8 minutes at −20°C), washed with phosphate-buffered saline (PBS), and blocked with 3% bovine serum albumin for 45 minutes at room temperature before antibody incubation. Primary antibodies directed against von Willebrand factor (VWF) (Abcam, #ab6994, 1:400) were incubated over night at 4°C. Samples were washed 3 times with PBS and subsequently incubated with appropriate metal-tagged secondary antibodies (¹⁷⁵Lu, Fluidigm #317002G) or fluorochrome-conjugated secondary antibodies (Alexa Fluor 488, Abcam) for 1 hour at room temperature. After washing with 0.005% Triton-X and PBS, nuclei were stained either with the Intercalator-^192/193Ir (Fluidigm, #201192A, 1:2000, 30 minutes) for LA-ICP-MS or DAPI (1 μg/mL, 5 minutes) for IF. After washing with ddH₂O, LA-ICP-MS samples were air-dried, whereas IF samples were washed with PBS and mounted for imaging.

For high-resolution LA-ICP-MS combined with immunohistochemistry, small ROIs (1 mm²) were ablated with a circular spot size of 5 μm, a fluence of 3 J/cm², and a scan speed of 40 μm/s. The ablated tissue was transported into the ICP-MS and analyzed as described previously. ¹⁵⁸Gd, ³¹P, and the metal isotopes of the secondary antibodies ¹⁷⁵Lu (endothelial marker) as well as ¹⁹²Ir (cell nuclei) were measured.

For immunofluorescent imaging, brain sections were imaged using a Zeiss Axio Observer.Z1 fluorescence microscope equipped with an AxioCam MRm camera and using a 10× objective and appropriate filter sets for DAPI and Alexa Fluor 488. Images were acquired using AxioVision software (AxioVision Rel 4.8).

Statistical Analysis

Statistical analyses were performed using SigmaPlot (Version 13.0; Systat Software Inc, San Jose, CA). The Kolmogorov-Smirnov test was used to assess Gaussian distribution. With the element concentrations from the individual ROIs, differences among the treatment groups were assessed by parametric t test or nonparametric Mann-Whitney U test, respectively, followed by Bonferroni's multiple comparison procedures, where appropriate. The degree of colocalization among the elements Gd, Fe, and Mn was assessed by correlation analysis of the measured element concentrations in the ROIs, using the coefficient of determination R² obtained by both Pearson (parametric) and Spearman (nonparametric) analysis as metrics. A P value <0.05 was regarded as significant.

RESULTS

A total of 42–45 slices of rat brains per group, each covering the brain in 1 of the 3 anatomical orientations, were analyzed using LA-ICP-MS, and the spatial distributions of Gd, Fe, Mn, and P—including quantification of the tissue concentrations of Gd, Fe, and Mn—were determined. Distinct patterns of spatial distribution of these elements were observed that corresponded to well-recognizable morphological structures of the rat brain.

Substantial Gd Concentrations After Gadodiamide Administration in Many More Brain Structures Than Previously Reported

After the injection of the linear GBCA gadodiamide, Gd was clearly visible in various distinct, but widely spread, structures of the CNS (Fig. 1). To estimate the amount of Gd in the brain, the average Gd concentration across all brain parts covered by an entire medial sagittal section including the white matter was determined. It amounted to 3.4 nmol Gd per gram of wet tissue. White matter as identified by high intensities in P imaging (vide infra) showed only minimal Gd. In the gray matter, several structures accumulating Gd were discernible, including structures previously reported to retain Gd after linear GBCA, but also in structures that were only now identified (Fig. 2). The average Gd concentration of the ROIs placed in distinct structures of the gray matter in accordance with the rat brain atlas³³ (n = 26; the OB and the choroid plexus [chp] are excluded for reasons detailed in the discussion [vide infra]) amounted to 17.2 ± 2.6 nmol/g (mean ± SEM; Table 1).

FIGURE 1.

Quantitative distribution of Gd in all 42 analyzed tissue slices of the rat hemispheres after gadodiamide in sagittal (top, from medial to lateral), horizontal (center, from dorsal to ventral), and coronal (bottom, from caudal to rostral) orientation. The color scale indicates the measured Gd concentration in nmol per g tissue.

FIGURE 2.

Quantitative distribution of Gd in the rat brain after gadodiamide in sagittal (A, medial; B, lateral), horizontal (C, medial; D, dorsal), and coronal (E–H, from caudal toward rostral) tissue slices. The color scale indicates the measured Gd concentration in nmol per g tissue. Arrows denote distinct structures of the brain. Images correspond to figures from Paxinos and Watson³³: A, Figure 172; B, Figure 174; C, Figure 202; E, Figure 126; F, Figure 111; G, Figure 83; H, Figure 61; D was not depicted. Abbreviations: DCN, deep cerebellar nuclei; cbw, cerebellar white matter; GrCb, granular layer of the cerebellum; MoCb, molecular layer of the cerebellum; pcuf, preculminate fissure; Ve, vestibular nucleus; SOC, superior olivary complex; 7N, facial nucleus; IC, inferior colliculus; SC, superior colliculus; SN, substantia nigra; Th, thalamic nuclei; Hth, hypothalamus; CPu, caudate putamen; GP, globus pallidus; RSC, retrosplenial cortex; DS, dorsal subiculum; Hi, hippocampus; S1, primary somatosensory cortex; Cg, cingulate cortex; OB, olfactory bulb; chp, choroid plexus (nonneuronal tissue lining the ventricles).

TABLE 1.

Elements Quantified in Distinct Brain Structures

		Gadodiamide			Gadobutrol			Saline
		Element Concentrations in, nmol/g
Structure	Ori	Gd	Fe	Mn	Gd	Fe	Mn	Fe	Mn
Entire medial sagittal section (whole-brain estimate)	Sag	3.4*	324.7*	9.7*	0.8†	398.7†	11.5†	495.7‡	10.9‡
Cerebellum
Deep cerebellar nuclei	Hor	18.2	494.8	12.3	0.4	676.7	20.6	817.8	13.4
	Sag	57.9	649.7	17.0	0.3	668.5	17.3	961.2	18.3
	Sag	32.2	734.4	19.5	0.6	885.5	19.7	863.7	9.4
	Sag	36.8	992.7	25.3	0.7	865.1	19.2	950.0	14.4
	Sag	31.3	685.0	15.2	0.7	1193.4	20.3	936.4	25.6
Granular layer of cerebellum	Sag	15.9	595.1	13.7	0.6	746.6	15.7	516.4	7.9
	Sag	21.0	626.4	13.2	0.5	821.8	15.1	702.1	15.4
	Sag	19.6	621.1	14.0	0.6	860.1	16.4	769.9	14.9
Brainstem
Vestibular nucleus	Sag	3.9	593.0	19.3	0.2	529.1	15.1	831.3	20.7
	Cor	3.1	593.1	18.8	0.5	924.1	24.4	799.3	23.1
Superior olivary complex	Sag	13.7	393.4	14.0	0.6	697.4	24.8	829.7	31.4
Facial nucleus	Sag	1.9	777.8	23.7	0.4	508.1	13.7	803.6	17.3
Midbrain
Inferior colliculus	Sag	7.3	833.6	25.1	0.8	941.8	24.7	842.0	22.6
Superior colliculus	Sag	4.7	460.7	15.9	0.6	621.0	16.5	591.0	16.3
Substantia nigra	Sag	10.5	652.1	11.4	0.7	897.2	13.1	558.7	8.3
Interbrain
Thalamic nuclei	Sag	5.7	444.6	14.8	0.6	497.3	15.7	515.4	14.9
Hypothalamus	Sag	4.6	410.7	14.0	0.5	415.7	11.9	412.3	13.7
Cerebrum
Caudate putamen	Sag	9.9	453.6	10.6	0.7	366.9	8.4	486.0	10.3
Retrosplenial cortex	Sag	12.2	457.4	10.6	0.7	543.0	14.5	486.8	11.6
	Cor	15.4	435.6	11.9	0.6	279.2	3.8	397.9	5.9
Dorsal subiculum	Sag	31.5	552.2	10.6	0.4	560.4	10.0	601.4	10.7
	Cor	20.4	538.2	10.5	0.4	293.4	1.9	515.9	4.7
Primary somatosensory cortex	Sag	6.2	404.6	13.0	1.4	404.6	8.9	469.5	8.3
	Hor	18.0	384.5	10.2	0.4	222.4	3.7	172.3	2.8
Cingulate cortex	Cor	10.0	338.7	9.0	1.3	320.0	6.1	349.6	9.5
	Hor	35.0	310.4	9.0	0.4	100.0	0.2	215.2	0.0
Separate structures§
Olfactory bulb	Sag	35.9	616.4	10.7	4.3	450.4	7.9	439.8	8.4
	Sag	14.0	463.5	10.4	2.7	620.0	9.1	478.1	9.4
Choroid plexus	Cor	11.6	870.7	8.7	15.8	1713.2	11.5	1208.2	8.7
	Sag	9.9	1594.7	7.5	15.5	734.0	6.3	1263.0	3.1
	Sag	10.4	1794.8	8.1	10.9	826.8	5.8	1690.4	9.4
Mean§		17.2	555.1	14.7	0.6	609.2	13.9	630.6	13.5
SEM§		2.6	32.0	0.9	0.1	52.5	1.4	44.4	1.4
Difference vs control		¶,∥	ns	ns	¶,∥	ns	ns	—	—
Difference vs other GBCA		¶	ns	ns	¶	ns	ns	—	—

Concentrations were measured by placing ROIs in individual tissue slices of the respective orientations.

*Average over 315,500 voxels.

^†Average over 278,200 voxels.

^‡Average over 309,400 voxels.

^§Separate structures excluded from arithmetic mean due to their unique properties (see Discussion).

^∥Gd concentrations in all ROIs of saline animals were below the limit of quantification.

^¶P < 0.0001, Wilcoxon signed rank test.

Ori, orientation; Sag, sagittal; Hor, horizontal; Cor, coronal.

In line with previous studies, the most prominent brain structures were the DCN with Gd concentrations of up to 57.9 nmol/g, the caudate putamen (CPu) including the GP with 9.9 nmol/g, the primary somatosensory cortex (S1) with up to 18.0 nmol/g, the OB with up to 35.9 nmol/g, and the chp with up to 11.6 nmol/g (Fig. 2; Table 1). Also, substantial amounts of Gd were measured in the granular layer of the cerebellum (GrCb), the cingulate cortex (Cg), the thalamus (Th), the hypothalamus (Hth), and certain parts of the hippocampus (Hi) (Fig. 2; Table 1).

Newly identified structures with high Gd concentrations included the retrosplenial cortex (RSC) with up to 15.4 nmol/g, the dorsal subiculum (DS) with up to 31.5 nmol/g, and some distinct layers of the Hi, all of which are situated closely together (Figs. 2, 3B; Table 1). Further newly identified structures comprise the superior olivary complex (SOC) with a Gd concentration of 13.7 nmol/g and the inferior colliculus (IC) with 7.3 nmol/g (Fig. 2; Table 1). The substantia nigra (SN), previously only reported for showing MR hyperintensity,^34,35 revealed a Gd concentration of 10.5 nmol/g (Fig. 2; Table 1).

FIGURE 3.

Details from images of Gd distributions after administration of gadodiamide in the cerebellum (A, horizontal orientation, detail of Fig. 3C; Paxinos and Watson Fig. 202), the hippocampus (B, sagittal orientation, detail of Fig. 3A; Paxinos and Watson Fig. 172), and the cortical forebrain (C, sagittal orientation, detail of Fig. 3B; Paxinos and Watson Fig. 174). The color scale indicates the measured Gd concentration in nmol per g tissue. Arrows denote distinct structures within the brain regions. Abbreviations: DCN, deep cerebellar nuclei; cbw, cerebellar white matter; GrCb, granular layer of the cerebellum; MoCb, molecular layer of the cerebellum; DS, dorsal subiculum; Or, oriens layer; Rad, radiatum layer; hif, hippocampal fissure; CA3, field CA3 of the hippocampus; GrDG, granular layer of the dentate gyrus; MoDG, molecular layer of the dentate gyrus; PoDG, polymorph layer of the dentate gyrus; S1FL, primary somatosensory cortex forelimb region; S1BF, primary somatosensory cortex barrel field; S1DZ, primary somatosensory cortex dysgranular zone; CPu, caudate putamen.

Images of Gd distribution at higher magnification revealed distinct spatial patterns of Gd after gadodiamide administration even within certain brain structures (Fig. 3). First, the GrCb showed high Gd concentration that increased toward the DCN, whereas Gd was virtually not present in the molecular layer (Fig. 3A). In the DCN, the Gd concentration increased laterally (Fig. 3A). Second, different parts of the Hi held different concentrations of Gd: the outermost oriens layer (Or) connected to the DS showed the highest local Gd concentrations, whereas the radiatum layer (Rad) displayed almost no Gd (Fig. 3B). In the dentate gyrus of the Hi (DG), the granular layer of the dentate gyrus (GrDG) contained Gd, whereas the molecular layer and the polymorph layer showed almost no Gd (Fig. 3B). Lastly, the S1 showed two circumscribed areas of high Gd, which are the forelimb region and the barrel field. The two are separated by the dysgranular zone with hardly any Gd (Fig. 3C).

Very Low Gd Concentrations After Gadobutrol Administration

After the injection of the macrocyclic GBCA gadobutrol, the average Gd concentration across all brain parts covered by a medial sagittal section including the white matter was 0.8 nmol/g. Averaged over the ROIs placed in distinct structures of the gray matter (n = 26; excluding the OB and the chp), Gd concentrations of rats receiving gadobutrol amounted to 0.6 ± 0.1 nmol/g. This is significantly lower than in rats receiving linear gadodiamide, amounting to only 3.5% of the latter (Table 1).

After the injection of gadobutrol, the Gd concentration was very low throughout the brain (Fig. 4). One exception was the chp, which was visually recognizable in all anatomical orientations with Gd concentrations up to 15.8 nmol/g (Fig. 5; Table 1). The second exception was the OB with Gd concentrations up to 4.3 nmol/g (Fig. 5; Table 1). Some Gd outside of the brain tissue borders can be seen in a few sections. These artifacts may originate from contamination in the Tissue-Tek embedding medium during tissue preparation.

FIGURE 4.

Quantitative distribution of Gd in all 43 analyzed tissue slices of the rat brain after gadobutrol in sagittal (top, from medial to lateral), horizontal (center, from dorsal to ventral), and coronal (bottom, from caudal to rostral) orientation. Tissue borders and ventricles in white were reproduced and transferred from the respective P images. The color scale indicates the measured Gd concentration in nmol per g tissue. Note: some artifacts are visible outside the tissue in the embedding matrix.

FIGURE 5.

Quantitative Gd distribution in the rat brain after gadobutrol in sagittal (A, medial; B, lateral), horizontal (C, medial; D, ventral), and coronal (E–H, caudal toward rostral) tissue slices. The color scale indicates the measured Gd concentration in nmol per g tissue. Arrows denote distinct structures of the brain. Tissue borders and ventricles in white were transferred from the respective P images. Images correspond to figures from Paxinos and Watson³²: A, Figure 168; B, Figure 175; C, Figure 204; E, Figure 136; F, Figure 108; G, Figure 83; H, Figure 70; D was not depicted. Note: some artifacts are visible outside the tissue in the embedding matrix. Abbreviations: OB, olfactory bulb; chp, choroid plexus (nonneuronal tissue lining the ventricles).

No Gd was detected in any brain sample of rats receiving saline solution, thus demonstrating that this treatment served as a successful negative control (Supplemental Digital Content, Fig. S2, http://links.lww.com/RLI/A887).

Distributions of P, Fe, and Mn

Phosphorus was found abundant in myelin lipid-rich regions such as the corpus callosum (cc), the anterior commissure (ac), the optic tract (opt), the lateral lemniscus (ll), and the cerebellar white matter (cbw), as well as regions known to have a considerable portion of white matter nerve tracts intermingled with nuclei such as the striatum (CPu) (Figs. 6A, B). In general, regions with high P content seemed to have low to nonexisting levels of Gd as well as Fe and Mn (Fig. 6).

FIGURE 6.

Distribution of P (A and B) and quantitative distribution of Fe (C and D) and Mn (E and F) in the rat brain after gadodiamide in sagittal (A, C, and E; Paxinos and Watson Fig. 172) and coronal (B, D, and F; Paxinos and Watson Fig. 83) tissue slices. The color scale for P indicates element levels in counts per second as detected in mass spectrometry. The color scales for Fe and Mn indicate the measured element concentration in nmol per g tissue. Arrows denote regions of the brain showing elevated levels of the respective element. Abbreviations: DCN, deep cerebellar nuclei; VeN, vestibular nucleus; 7N, facial nucleus; SOC, superior olivary complex; IC, inferior colliculus; SN, substantia nigra; IP, interpeduncular nucleus; GrDG, granular layer of the dentate gyrus; CPu, caudate putamen; cbw, cerebellar white matter; ll, lateral lemniscus; cc, corpus callosum; ac, anterior commissure; opt, optic tract; dhc, dorsal hippocampal commissure; dcw, deep cerebral white matter; cp, cerebral peduncle; OB, olfactory bulb; a, artery.

Mean concentrations of Fe and Mn in the gadodiamide, gadobutrol, and saline groups, as averaged over the ROIs (n = 26 per group; excluding the OB and the chp), revealed no significant differences among the groups (Table 1).

Across all 3 treatment groups, the highest concentrations of Fe were observed in arteries, veins, and blood capillaries of the chp (Figs. 6C, D; Table 1). Relatively high Fe concentrations were measured in gray matter structures, such as the DCN, IC, GrDG, SOC, the facial nucleus (7N), SN, the interpeduncular nucleus (IP), and the OB (Figs. 6C, D; Table 1).

Compared with Fe, Mn was detected at considerably lower levels and showed a more even distribution with only few regions of relatively high concentrations, so that morphological structures were not easily discernible in the Mn images (Figs. 6E, F; Table 1). The difference between gray and white matter was not as concise as for Gd and Fe. Relatively high concentrations of Mn were measured in the IC, SOC, and the cerebellum, where the DCN and the GrCb showed a higher concentration than the cbw (Figs. 6E and F; Table 1).

Colocalizations of Elements: No Significant Correlations of Gd With Fe and Mn

The overlay of Fe and Gd images from rats receiving gadodiamide indicated that the 2 elements barely colocalize. The exceptions were the OB, DCN, and IC, where some colocalizations were observed (Fig. 7B). Fe and Gd images at higher magnification demonstrated distinct spatial patterns for both elements, which showed some but no distinct colocalizations in areas and layers of the cerebellum, the Hi, and the cerebellar cortex (Supplemental Digital Content, Figs. S3 and S4, http://links.lww.com/RLI/A887). After the administration of gadobutrol, no colocalizations of Gd and Fe were apparent, apart from the chp (Supplemental Digital Content, Fig. S5, http://links.lww.com/RLI/A887).

FIGURE 7.

Overlays of Gd in red together with P (A), Fe (B), and Mn (C), respectively, in green after gadodiamide in a single sagittal tissue slice (Paxinos and Watson Fig. 172). A shift in color toward yellow indicates the presence of both elements. Arrows denote regions of the brain showing elevated levels of both elements. Abbreviations: cbw, cerebellar white matter; ll, lateral lemniscus; cc, corpus callosum; CPu, caudate putamen; ac, anterior commissure; opt, optic tract; DCN, deep cerebellar nuclei; SOC, superior olivary complex; IC, inferior colliculus; SN, substantia nigra; OB, olfactory bulb; chp, choroid plexus (nonneuronal tissue lining the ventricles).

Correlation analysis of the concentrations of Fe and Gd for the individual ROIs (n = 28 per group; excluding chp) revealed that there is no significant correlation between these elements (Fig. 8) for both the gadodiamide group (Pearson R_p² = 0.08; Spearman R_s² = 0.06; Spearman p_s = 0.22) and the gadobutrol group (R_p² = 0.01; R_s² = 0.03; p_s = 0.40).

FIGURE 8.

Correlation analysis between Gd and Fe (top), Gd and Mn (center), as well as Fe and Mn (bottom) after administrations of gadodiamide (red), gadobutrol (blue), and saline solution (green). The coefficient of determination for Pearson (R_p²) as well as Spearman (R_s²) is listed together with the Spearman (nonparametric) P value (p_s). Analysis excluded concentrations for the choroid plexus.

The overlays of Mn and Gd images after the administration of gadodiamide showed some colocalization only in the SOC, in the DCN, and in the IC (Fig. 7C).

Correlation analysis of the concentrations of Mn and Gd for the ROIs (n = 28 per group; excluding chp) revealed that there is no significant correlation between these elements (Fig. 8) for both the gadodiamide group (R_p² < 0.01; R_s² = 0.04; p_s = 0.31) and the gadobutrol group (R_p² = 0.05; R_s² < 0.01; p_s = 0.95).

Interestingly, analysis for colocalization of Fe and Mn for the ROIs (n = 28 per group; excluding chp) revealed significant correlations with moderate coefficients of determination for all 3 treatment groups (Fig. 8): gadodiamide group (R_p² = 0.63; R_s² = 0.43; p_s < 0.001), gadobutrol group (R_p² = 0.70; R_s² = 0.72; p_s < 0.001), and saline group (R_p² = 0.58; R_s² = 0.52; p_s < 0.001).

Colocalization With the Endothelium

We further explored the DCN regions for possible colocalization with cellular markers using immunohistochemistry high-resolution LA-ICP-MS.

The approach was validated by comparing the staining patterns generated by fluorochrome-labeled secondary antibodies and metal-tagged secondary antibodies for the endothelial marker VWF (¹⁷⁵Lu) as well as the cell nuclei stain with ¹⁹²Ir. Both approaches gave very similar staining patterns for endothelium and cell nuclei in consecutive tissue sections, as shown in a region of the brain stem (Supplemental Digital Content, Fig. S1, http://links.lww.com/RLI/A887).

For visualization of the Gd distribution, combinations of the Gd images with the images of the cellular markers were false colored and overlaid to show the spatial correlation of Gd, VWF, and cell nuclei.

The concentration of Gd remained high in the DCN after the washing and staining procedures, indicating that the remaining Gd is most likely tissue bound or has formed insoluble precipitate. The overall Gd pattern in the DCN clearly outnumbered the observed vascular structures, only partially overlapped with the endothelial marker VWF and contained additional spots that were devoid of the marker VWF (Fig. 9). The nuclei pattern was homogenously distributed except for a dense nuclei staining in the granular layer and only showed minor overlap with Gd, mostly in areas that were additionally positive for VWF.

FIGURE 9.

Distribution of Gd, the endothelial marker von Willebrand factor, and cell nuclei in the deep cerebellar nuclei identified by immunohistological LA-ICP-MS after multiple administrations of gadodiamide and gadobutrol. Shown are caudal coronal tissue slices (Paxinos and Watson Fig. 126). Representative pseudo-colored overlays of Gd (red), von Willebrand Factor (VWF, green), and an iridium containing nucleic acid intercalator (blue) are shown at 3 different magnifications. Fields of higher magnification are indicated by squares.

For gadobutrol, only few Gd spots remained in the brain slices after the washing and staining procedure. The substantial loss during washing indicates that Gd is present in its highly water-soluble chelated form. Moreover, the remaining Gd spots were positive for VWF showing localization to the vasculature (Fig. 9).

DISCUSSION

The present study provides a comprehensive quantitative analysis of the spatial distribution of Gd and endogenous Fe and Mn in the whole brains of healthy rats after administration of high cumulative doses of GBCA. Our study corroborated high Gd concentrations after gadodiamide in previously reported distinct structures of the CNS, but also identified novel structures with high Gd concentration. Facilitated by the analysis of brain sections in 3 orientations, Gd presence was even detected in circumscribed subregions and layers of certain brain structures that differ regarding their cytoarchitecture and function. The identification of distinct spatial patterns of residual Gd may pave the way for tailored functional neurological testing. Significantly lower concentrations of Gd were measured after the administration of the macrocyclic GBCA gadobutrol than after the linear GBCA gadodiamide. Our quantitative analysis exhibited no significant colocalization of Gd and Fe as well Gd and Mn. This result argues against a general role of these metals in local transmetalation and release of Gd³⁺ ions in the CNS.

Gadolinium Presence in Numerous Brain Regions After Gadodiamide Injection

Although the elemental Gd images clearly showed widespread distribution of Gd in the brain, the fraction of the injected amount of the linear GBCA gadodiamide present in the rat brain 5 weeks after the last injection was very small. The average Gd concentration measured across all brain parts covered by a medial sagittal section (including the OB and the chp) was 3.4 nmol Gd per gram of tissue. Extrapolating this concentration to the mass of the whole brain,³⁶ this amounts to only 0.00024% of the administered gadodiamide, which is in line with previous reports.²⁶

After the injection of gadodiamide, Gd was broadly distributed throughout the brain, yet distinct morphologic structures with prominent Gd presence were clearly discernable. In line with previous reports, Gd was largely absent from the white matter as determined by the spatial distribution of high P content. This can be ascribed to the high polarity of GBCAs in contrast to the low polarity of phospholipid-rich myelin of nerve fibers present in the white matter.³⁷ Studies focusing on the chemical species of Gd present in the brain found a portion of Gd after injection of linear GBCA retained as insoluble species.^26,38 Electron microscopy and energy-dispersive x-ray spectroscopy revealed amorphous, spheroid structures of 100–200 nm that were most probably composed of mixed Ca-/Gd-phosphate salts.^21,38 Yet, the spatial resolution of LA-ICP-MS does not enable us to identify such small foci of Gd-P colocalization.

In the gray matter, we identified several structures with high Gd concentration. We measured some of the highest Gd concentrations in ROIs of the very regions, which have originally led to the discovery of Gd uptake after linear GBCA by showing signal hyperintensities in MRI of human brains, namely, the DCN, GP, and CPu.^1,2,4,39,40 Inductively coupled plasma mass spectroscopy studies corroborated these findings,^{6,22,26,41–44} but also discovered additional brain structures with residual Gd, such as the OB, the Th, and cerebellar gray matter.^21,24,42 Studies applying LA-ICP-MS to selected tissue slices of some CNS portions revealed additional regions and layers accumulating Gd such as the thalamic nuclei and the GrCb.^21,22,24,25 Our results qualitatively corroborate these previously described structures with prominent Gd content, and the Gd concentrations we measured largely agree with the concentration ranges previously reported from ICP-MS and LA-ICP-MS studies.^{21,22,24–26}

More recently, Minaeva and colleagues²³ used LA-ICP-MS to study the spatial distribution of Gd in serial coronal sections of whole rat brains after administration of the linear GBCA gadopentetate. The authors identified Gd uptake in several regions that had hitherto not been reported: the Cg, the piriform cortex, fields of the Hi and the DG, the facial nucleus (7N), and the vestibular nucleus. They also generated Gd distribution maps of the cerebral cortex in brain samples from GBCA-exposed deceased patients and found Gd in distinct cortical layers. In a subsequent study, these authors additionally reported an uptake of Gd in the S1.²⁵

By generating numerous sequential sections in all 3 spatial orientations—coronal, sagittal, and horizontal—throughout the rat brain, we were able to provide a more comprehensive quantitative analysis of the spatial distribution of Gd. Although we qualitatively corroborated most—but not all—of the results obtained by Minaeva et al²³ and Hua et al,²⁵ we also identified new structures with high Gd concentrations. Moreover, we discovered that, even within certain structures, circumscribed anatomical subregions and/or distinct histological layers contained plainly different levels of Gd.

We discovered high Gd levels in the retrosplenial cortex (RSC) and in the dorsal but not the ventral subiculum of the Hi. In the SN, we measured a high Gd content not previously reported but in line with reports of MR hyperintensity after linear GBCA in humans.^34,35 Further, we measured relatively high Gd concentrations in the inferior colliculus (IC) but only low Gd concentrations in the superior colliculus (SC). This observation is consistent with differing functions of the two: the IC is part of the auditory system, whereas the SC serves visual processing. Increased Gd in and a potential effect on the IC would be a possible explanation for the observed transient decrease in the acoustic startle response of rats after the injection of linear GBCA recently reported.³²

Minaeva et al²³ reported very high Gd content after linear GBCA in the 7N and the vestibular nucleus located in the pons. The 7N contained the highest Gd concentration of all ROIs assessed by these authors.²³ In our study, however, the Gd concentration in the area of the 7N in all 3 spatial orientations was very low (1.9 nmol/g), ranking it one of the lowest ROIs. A similar discrepancy was observed for the vestibular nucleus. Intriguingly, in close proximity to the 7N lies the SOC,³³ for which Gd presence has not been previously described. We measured a high Gd concentration of 13.7 nmol/g in the SOC, which is also a major part of the auditory system.⁴⁵

We detected major differences in the Gd concentrations among specific histological parts of the Hi,³³ namely, high concentration in the oriens layer (Or) compared with virtually no Gd in the radiatum layer (Rad). Even within the DG, different Gd concentrations were observed for its specific histological layers that differ with regard to their cytoarchitecture.⁴⁶

Although we measured some of the highest Gd concentrations in the S1, we discovered that Gd was not homogeneously distributed within this structure. In both sagittal and coronal orientation, we identified two circumscribed areas of high Gd presence separated by a third area with hardly any Gd. These areas correspond to the forelimb region, the barrel field, and the dysgranular zone, respectively, which differ regarding their cytoarchitecture, somatotopic representation, and functions.^47,48

In accordance with earlier reports, we observed high Gd concentrations in the chp.^23,25 The chp lines the ventricles of the brain and constitutes the blood–cerebrospinal fluid (CSF) barrier. It consists of blood vessels, connective tissue, and epithelium, and produces and secretes the CSF.^49,50 As the endothelium of its capillaries is fenestrated, small hydrophilic molecules including GBCAs can pass through. In fact, along with involvement of the blood-brain barrier and the glymphatic system, passage through the blood-CSF barrier is suggested to be the central pathway for Gd species from the blood into the brain.^27,51

The very high Gd concentration of the OB can at least in part be ascribed to additional and unique transport pathways. The OB is connected with the nasal epithelium by the olfactory nerves as well as by blood and lymphatic vessels that pass through the fenestrae of the cribriform plate that separates the nasal cavity from the cranial cavity.⁵² Because the flow of CSF along the olfactory nerves into the nasal epithelium is a major pathway of CSF drainage and the interface between the nerves and the bulb is a brain region with high CSF flow,⁵² the bulb would be particularly exposed to GBCA contained in the CSF. Moreover, GBCA may well reach the OB via the blood vessels passing through the cribriform plate.⁵²

Recognizing the remarkable spatial distribution of Gd throughout the brain poses the question as to why some regions accumulate considerable amounts of Gd while others show hardly any Gd. Our finding of marked differences in the Gd concentrations of specific regions and even among layers within given regions may reflect different susceptibility to Gd uptake based on the differences in their respective cytoarchitecture. A case in point is the S1 displaying high Gd concentrations in remarkable contrast to all of the adjacent areas of the cerebral cortex, which showed only negligible Gd concentrations.⁴⁷ Moreover, this also applies to specific fields of the S1, which may explain the large differences of Gd concentration detected in the forelimb region, the barrel field, and the dysgranular zone.^48,53 Similarly, differences in the cytoarchitecture may be the reason for differences observed among specific histological parts of the Hi and within the cerebellar gray matter.^46,54

Besides cytoarchitectural differences, different neuronal activity of brain structures might have an influence on their susceptibility to Gd uptake. Habermeyer and colleagues³² found that synaptosomes contain small amounts of Gd in rats, particularly after linear GBCA. As the frequency of synaptic transmission, in general, goes along with neuronal activity, higher activity may result in higher synaptosomal uptake of Gd. Indeed, a study showed different flow speeds of the interstitial fluid in different brain regions and a marked reduction of transport of a GBCA in the brain interstitial space after neuronal excitation.⁵⁵

The dissociation of the Gd³⁺ ion from its contrast agent ligand (dechelation) and its subsequent deposition in several tissues including the CNS is probably facilitated by transmetalation and by the presence of Gd acceptors. The competition between endogenous metal ions, such as Fe, Mn, Cu, and Zn ions, and the Gd ion for the contrast agent ligand has been proposed to play a major role in Gd uptake and deposition.^{12,42,56–58} Endogenous Gd acceptors, such as macromolecules and inorganic phosphate or carbonate, may compete with the contrast agent ligand for the Gd ion.^21,38,59 Transmetalation and Gd binding to endogenous acceptors are probably closely connected, because the free Gd ion will most likely not exist for a longer time in the physiological environment where the acceptors are readily available at up to millimolar concentrations. Gd ions bound to macromolecules or inorganic anions may be precipitated as nanoparticles in the tissue.^21,38 However, where and when these processes takes place is not known, and no definitive information is available in which chemical form the Gd is present in tissue, most likely it is not the injected gadodiamide.^{12,21,23,24,26,27,38,42} These different chemical species may be a further reason for selective uptake or deposition of Gd in specific brain regions. The investigation of the distribution of endogenous metals and potential colocalization with Gd was therefore an important part of this study. Macrocyclic GBCAs, such as gadobutrol, are much less affected by dechelation due to their high kinetic inertness.¹²

Colocalization of Gadolinium With Iron, Manganese, and Endothelium

The very first regions identified to contain Gd after linear GBCA administration—DCN, GP, and striatum—have long been recognized to be rich in Fe.^28–30 In the face of the pronounced Gd presence in Fe-rich brain structures, the notion of transmetalation as a major contributor to Gd accumulation in the CNS seemed quite plausible.^12,23,27 Reports on Gd-Fe colocalizations as derived from visual inspections of juxtaposed or overlaid element maps obtained by LA-ICP-MS seem to support this notion.^21,23–25 One study analyzed selected homogenized brain portions and found correlations between Gd and Fe concentrations after administration of linear GBCAs.⁴² However, the rats in this study were subtotally nephrectomized, which impairs renal elimination of the GBCA, and moreover causes severe urinary Fe loss.⁶⁰

From visual inspection of Gd-Fe overlay images after gadodiamide, a rather weak colocalization could be surmised in the OB, in the DCN, and in the IC. On the other hand, distinctive distribution patterns of both Fe and Gd were determined, respectively, but found not to be congruent, for example, in the Hi and the GrCb. To create clarity, we tested the statistical correlation between Gd and Fe concentrations measured in the 26 ROIs (excluding the OB and the chp). This analysis clearly demonstrates that there is no statistical correlation between Gd and Fe after the injection of gadodiamide. Likewise, no statistical correlation between Gd and Mn was found. This result argues against a general role of these metals in local transmetalation and release of Gd³⁺ ions in the CNS.^12,23,27

In addition, LA-ICP-MS does not differentiate among available Fe species, for example, heme iron, ferritin, or transferrin, present in different brain structures. The majority of iron is bound to such high affinity carriers, and only traces are present as free Fe³⁺ ion, which could initiate the transmetalation pathway. Thus, further studies are necessary to understand possible interactions between Gd and distinct iron species.⁶¹

By investigating the localization of Gd and the endothelial marker VWF, we showed that a substantial amount of Gd is present outside the vasculature and thus is not associated with blood-derived iron. It needs to be elucidated whether these Gd spots might correlate with other markers (eg, neuronal) or localize to the interstitial space. Moreover, the Gd pattern was unchanged after the fixation, washing, and staining procedure, indicating that the retained Gd is bound to endogenous molecules or present as insoluble precipitates, which is in line with the current knowledge on Gd species present in the brain after administration of gadodiamide.^26,38

Significantly Less Gadolinium Presence After Gadobutrol

Extrapolating the average concentration from a whole medial sagittal section in the same manner as for gadodiamide, only 0.00006% of Gd administered as gadobutrol was detected in the brain, which is 4.3 times less than for gadodiamide, in line with previous reports.²⁶ With the exception of the OB and the chp, infinitesimally low concentrations of Gd from gadobutrol were detected. Accordingly, the Gd concentration averaged over the ROIs placed in distinct structures of the gray matter (n = 26; excluding the OB and the chp) was significantly lower after gadobutrol (0.6 ± 0.1 nmol/g) than after gadodiamide (17.2 ± 2.6 nmol/g). This is in line with the vast majority of clinical and preclinical studies by both MRI and elemental analysis.^{21,22,24,25,27} The lower tissue concentration observed after gadobutrol injection reflects its much higher kinetic inertness toward release of free Gd³⁺ ions.^11,57

Two structures, the OB and the chp, had Gd concentrations similar to those after gadodiamide. This may again be ascribed to the unique characteristics of these structures as detailed previously for gadodiamide.

Moreover, the Gd pattern observed after administration of gadobutrol is different and seems to be localized in close proximity to the ventricles and larger vessels. Indeed, visual inspection of Gd-Fe overlay images after injection of gadobutrol revealed partial colocalization with Fe-rich vascular structures, such as the chp. However, as for gadodiamide, there was no general colocalization with Fe or Mn across the brain. Statistical correlation analysis for all 26 ROIs (excluding the OB and the chp) clearly demonstrates that there is no statistical correlation between Gd and Fe, as well as between Gd and Mn.

In line with the observed Gd pattern, our immunohistochemistry results confirm the proximity of Gd to the vasculature after administration of gadobutrol. Moreover, the substantial reduction of Gd spots after the washing and staining procedure indicates that gadobutrol is present as intact chelate, which is washed out during the experimental procedure due to its hydrophilic nature. This is in line with previous data that only observed intact gadobutrol in brain homogenates.²⁶

Limitations of the Study

The present study has some limitations. We aimed to map Gd, Fe, Mn, and P throughout all parts of the rat brain using serial sections in all 3 spatial orientations. To achieve this, a total of 130 tissue slices were analyzed with a pixel resolution of 11.5 × 50 μm². Although previous studies measured these elements in homogenized brain section at several time points after GBCA administration,^22,62 our analysis involved only 1 time point, due to the very long measuring times. For the same reason, we also limited the number of rats per group to 3, with 1 rat per group for each orientation. Some variability in the concentrations measured in the different orientations may thus reflect interindividual variations. With a slice thickness of 10 μm and relatively large distances between analyzed slices, many structures were only visible in a single slice and thus the measured concentrations may not reflect the average concentrations in the entire structure. Similarly, the estimated Gd content for the whole brain may not be exact, originating from 1 medial sagittal tissue slice. However, the calculated remaining fraction of Gd is in line with previous reports. Although the applied resolution was sufficient to even delineate the Gd distribution within certain brain structures, it did not permit the identification of cellular structures without additional methodology. In a first attempt, we were able to visualize the Gd localization and endothelial structures, but other cell types, for example, neurons, have not been explored so far. Lastly, we studied one linear GBCA and one macrocyclic GBCA, yet conclusions about other GBCAs of the two classes should be drawn with due caution.²²

CONCLUSIONS

The study provides for the first time a comprehensive quantitative analysis of the spatial distribution of retained Gd as well as of endogenous Fe and Mn of the entire rat brain in all anatomical orientations after high doses of gadodiamide or gadobutrol. In rats exposed to gadodiamide but not in rats exposed to gadobutrol, many distinct structures, including specific subregions and layers with different cytoarchitecture and functions, throughout the gray matter were identified that were not previously reported to contain Gd. These results may pave the way for tailored functional neurological testing for potential adverse effects, as exemplified for structures of the auditory system and distinct areas of the S1. The results also indicate that local transmetalation with endogenous Fe or Mn is unlikely to explain the spatial patterns of Gd deposition in the brain, which argues against a general role of these metals in local transmetalation and release of Gd ions in the CNS. To further elucidate the mechanism of Gd uptake, differences among brain structures regarding their (molecular) cytoarchitecture and/or neuronal activity should be addressed.