Introduction

Metals play an important role in diverse biochemical and physiological functions in humans, and various enzymes and vitamins contain metals. Several metals are essential nutrients in humans, and metal shortages lead to a variety of deficiency diseases or syndromes. On the other hand, excess amounts of such metals induce cellular and tissue damage, leading to a variety of adverse effects and diseases. The appropriate concentration of a metal lies within a very narrow range, and any excess or shortage easily causes toxic effects in vivo [1–3]. The method for quantifying the metal content in affected organs is important not only for diagnostic purposes but also for the assessment of disease progression and treatment effects. Magnetic resonance imaging (MRI) is the most potent and sensitive technique for detecting abnormal metal deposition; this technique involves evaluating small changes in relaxation times [4, 5]. Most substances in the body are diamagnetic, and thus barely influence magnetic fields. On the other hand, most metals are classified as paramagnetic, and show magnetization only in the presence of a surrounding magnetic field. Paramagnetism influences the T1 and T2 relaxation times due to the influence of unpaired electrons. Some metals, such as nonionic nickel, iron, and gadolinium (at temperatures below 18 °C), are ferromagnetic, and present very strong magnetization that causes metal artifacts in MRI. However, ferromagnetic metals do not exist in the body, except in surgical or traumatic implants [6]. Most metal deposition causes high signal intensity on T1-weighted images (T1WI) [7, 8] and low signal intensity on T2-weighted images (T2WI) [9], and these signal intensity changes are rather specific for metal deposition in vivo. This article focuses on the diseases associated with metal deposition in the brain and metal detectability on brain MRI.

Difficulties associated with metal analysis in the brain

Several factors make metal analysis uncertain. First, brain tissue from patients is not easily acquired. Many studies of metal deposition have based diagnoses on clinical or image analysis, while only a small number of studies have relied on histological analysis. However, except for brain tissue analysis, the cause of the abnormal signal intensity on MRI cannot be determined accurately [5]. Second, excess concentrations of multiple metals are often observed in the same subject. For example, patients with metal poisoning from exposure in mines frequently have excess deposition of multiple metals, since metal ores usually include multiple kinds of metals. In addition, abnormal metal storage in the human body occasionally affects the absorption of other metals. Iron deficiency induces enhanced absorption of iron mediated by greater expression of metal transporters of iron, as well as the increased absorption of other metals such as manganese, nickel, lead, and/or cadmium [10–13]. Third, the magnetism of a metal varies according to the oxidation state and the type of compound involved, and its signal intensity on MRI varies according to these parameters and the status of the metal. The oxidation state of a metal and the compound it is in frequently change in the body, so the MRI signal intensity of a metal is not constant [14, 15]. Below we review the effects of metals on brain MRIs in order of the atomic number of the metal.

Calcium

Calcium deposition in the brain occurs with Fahr disease, various metabolic diseases, and tumors [5]. Calcification is easily detected on CT, but it is difficult to detect on MRI because of the variety of signal intensities produced by calcified regions [16]. This variety of signal intensities from calcification is not caused by its paramagnetic effect but by a surface effect whereby the irregular surface of calcium restricts the motion of water close to the Larmor frequency, with the T1 relaxation time of water shortening as the protein content in the water increases. The surface effect varies according to the surface structure of the calcification, so the signal intensity from calcification on T1WI can vary even when there is the same concentration of calcium [17, 18]. On T1WI, the signal intensity of calcification is strongest at 30 % calcium concentration by weight, while on T2* the signal intensity and calcium concentration show a linear correlation [18, 19]. However, these results were obtained in vitro, and the true mechanism of the high signal intensity from the calcified region has not yet been determined because calcium tends to accumulate in the basal ganglia, where other paramagnetic metals such as iron and manganese also accumulate. Calcium causes a high signal intensity on T1WI due to the surface effect in vitro, but the cause of the high signal intensity from calcified regions in vivo is unclear.

Manganese

Manganese neurotoxicity was first reported by Couper [20] in 1837 in a case series of 5 manganese miners. Patients with manganese neurotoxicity suffer from Parkinson’s symptoms, including tremor in the extremities, gait disturbance, and whispering speech [20–22]. Manganese is highly paramagnetic, and MRI is expected to detect its presence in the brain. The first study used monkeys that were administered MnCl2 by intravenous injection or inhaled aerosol. Regardless of the administration route of MnCl2, an increasing signal intensity on MRI is observed in the caudate nucleus, globus pallidus, substantia nigra, ventromedial hypothalamus, and pituitary gland [23]. Even in man, a hyperintense globus pallidus was found in patients receiving long-term total parenteral nutrition therapy that included manganese [24] and in workers exposed to manganese [25–27], with these hyperintensities diminishing after cessation of the manganese exposure [25, 28, 29]. Intravenous administration or inhalation of manganese causes manganese deposition in the brain, but the oral intake of manganese rarely results in manganese deposition. Intellectual impairment in children after excess oral intake of manganese has been reported, albeit without histological proof [30]. Most orally ingested manganese does not remain in the human body, due to several mechanisms. Manganese is present in food and water, but the intestine adjusts manganese absorption and only 1–5 % of the manganese ingested is absorbed [31]. In addition, the absorbed excess manganese is promptly excreted into the bile from the liver and rarely causes manganese toxicity [22, 32]. Iron deficiency anemia promotes manganese absorption in the intestine and elevation of the manganese concentration in the blood, but the signal intensity of the globus pallidus on T1WI is not affected [33]. In this way, manganese neurotoxicity is caused by mine dust inhalation or intravenous administration but rarely by oral intake [34, 35]. The manganese metabolic pathway adapts to oral MRI contrast agents containing manganese, and the contrast effect of this oral contrast agent based on manganese is limited to liver and intestine [36, 37].

Patients with chronic hepatic failure also show a hyperintense globus pallidus and to a lesser extent putamen and red nucleus on T1WI (Fig. 1) [38, 39]. The pattern of MRI abnormalities resembles that of manganese overload. Patients with liver cirrhosis and high signal intensity in the globus pallidus on T1WI have higher blood manganese concentrations than healthy volunteers [40]. The hyperintense globus pallidus on T1WI seen in patients with chronic hepatic failure due to Alagille syndrome diminished 3 months after hepatic transplantation. A high manganese concentration in the blood and symptomatic manganese toxicity were reported to improve following hepatic transplantation [41]. Manganese is considered to be the cause of hyperintense globus pallidus in patients with chronic hepatic failure, but copper and iron as well as manganese concentrations are elevated in the globus pallidus of autopsy cases [42]. The hyperintense globus pallidus seen in hepatic failure occurs due to an elevated concentration of manganese in the blood, which passes through the portal-systemic collateral vessels without migrating through the liver [43]. The hyperintense globus pallidus observed in schistosomiasis [44], Osler–Weber–Rendu disease [45], and post-transjugular intrahepatic portosystemic shunt placement [46] is also considered to be related to portal-systemic shunting. On the other hand, a hyperintense globus pallidus sometimes develops in patients with chronic liver failure in the absence of portal-systemic shunting [47]. This is because liver dysfunction associated with insufficient biliary excretion of manganese leads to an increased manganese concentration in the blood and hyperintensity of the globus pallidus [48].

Fig. 1
figure 1

A 57-year-old man with liver cirrhosis. A hyperintense lesion is present in the globus pallidus (white arrow) on sagittal T1WI due to manganese deposition

Full size image

Transient hyperintensity of the striatum on T1WI after an ischemic event is also associated with the manganese concentration. In a study on brief focal ischemia in rats, transient hyperintensity of the striatum was observed from 5 days to 4 weeks after the ischemia, as was a synchronous increase in the manganese concentration in the striatum. The increased manganese was due to increases in mitochondrial manganese superoxide dismutase and glutamine synthetase, which are manganese-containing metalloproteins. Endogenous manganese also can affect the signal intensity from the brain [49].

Iron

The iron distribution in the human brain was first uncovered by histochemical analysis (Prussian blue or Perls’ stain). However, more recently, biochemical analytical techniques such as atomic absorption spectrophotometry (AAS), inductively coupled plasma mass spectrometry (ICP-MS), or X-ray fluorescence spectrometry (XRF) has been used to evaluate the iron concentration in brain tissue owing to the precise and quantitative nature of the data obtained [4, 50]. The iron in the brain is divided into heme iron and non-heme iron. Heme iron is present in hemoglobin, whereas the non-heme iron includes low-molecular-weight complexes, metalloproteins such as transferrin, storage proteins such as ferritin and hemosiderin, and ionic iron. Ferritin and hemosiderin are the most important sources of iron-related signal changes in cerebral MRI, since ferritin and hemosiderin contain 33–88 % of the non-heme iron found in the brain [4, 50]. Ferritin, which is the main chemical form of iron in the brain, has a T2-shortening effect and shows a low signal intensity on T2WI. More than 0.01 mg/ml of iron can be detected on MRI in the laboratory [51], and the iron concentration in the human brain is up to 0.2 (mg/g tissue wet weight) in the globus pallidus, which is theoretically sufficient to evaluate the iron concentration with T2WI [50]. The iron concentration in the basal ganglia in the brain increases until 30 years of age on MRI [52] (Fig. 2). However, the iron concentration estimated via T2WI is not well correlated with the results of biochemical analysis [53, 54]. There are several reasons that the iron content estimated with T2WI is not accurate. One is that the regional T2 signal intensity is affected not only by iron but also by the water and myelin contents [4]. The second is that iron transforms various chemical structures in vivo, causing different paramagnetic effects. For example, oxyhemoglobin and deoxyhemoglobin both contain an Fe2+ species, but because of differences in their electronic structures, oxyhemoglobin is diamagnetic while deoxyhemoglobin is paramagnetic. However, apart from ferritin, which has been extensively studied, little is known about the magnetic properties of other forms of iron and their contributions to the MRI contrast [4]. The third reason is the clustering effect of iron particles. Even in the same chemical structures, the heterogeneous distribution of iron particles causes a T2-shortening effect (clustering effect) and a low signal intensity on T2WI [55–58]. As the correlation between the iron content and T2WI is questionable, T2*-weighted imaging [59], susceptibility-weighted imaging [60], and quantitative susceptibility mapping [61, 62] have been tested as methods of estimating the iron content. While the iron content can indeed by estimated more accurately with MRI using these techniques, the paramagnetic effects caused by other elements still affect such measurements. In addition, hemosiderin shows the best image contrast on T2*-weighted images and susceptibility-weighted images [63, 64], but ferritin shows more prominent image contrast on T2-weighted images obtained by conventional spin-echo images than on T2*-weighted ones [63].

Fig. 2
figure 2

A 15-year-old female (a) and 43-year-old female (b), both without any past history of severe disease. Due to iron deposition, the signal intensity of the globus pallidus (white arrows) is lower in the former than in the latter on T2WI

Full size image

Although various factors other than iron content also affect the estimated iron concentration in the brain on MRI, MRI is used to diagnose many kinds of iron-storage diseases, such as neurodegeneration with brain iron accumulation (NBIA) [65–68], Alzheimer’s disease [69, 70], Parkinson’s disease [71, 72], multiple sclerosis [73, 74], Friedreich’s ataxia [75, 76], and Huntington’s disease [53, 77]. NBIA is a group of neurodegenerative diseases characterized by iron accumulation in the basal ganglia. The diagnosis of NBIA has advanced with the development of MRI, gene analysis, and postmortem histological analysis. Ten disease-associated genes have been identified so far, but 20 % of NBIA cases remain genetically undefined. Only two of these NBIA genes are associated with iron homeostasis; the others are associated with mitochondria or fatty acid metabolism [66]. All of the NBIA disorders feature iron deposition in the globus pallidus but differ in their other findings. They are discriminated based on clinical movement disorder and neurodegenerative course [64]. In other iron-storage diseases, the true mechanism underlying the iron deposition is not known.

Iron has a T1-shortening effect, and its usefulness as a positive contrast agent for brain MRI [ultrasmall superparamagnetic iron oxides (USPIO)] has been examined. The signal intensity of iron on T1WI is influenced by the size of the iron particles [78], the clustering effect [55, 56], and T1WI parameters [79]. However, compared with gadolinium, the pattern of iron enhancement is heterogeneous and variable, so iron-based contrast agents cannot be used instead of gadolinium-based contrast agents (GBCAs) [80, 81]. The T1-shortening effect of iron is seldom used for the quantitative analysis of iron; instead, it is used to qualitatively check for the presence of iron. Subacute hemorrhage and melanin are typically identified by the presence of hyperintensity on T1WI [8]. The hyperintensity seen in subacute hemorrhage on T1WI is due to methemoglobin, which contains Fe3+. Even in artificial cerebral spinal fluid, hemoglobin changes to methemoglobin after incubation for 84 h at 38 ° C, just like the chemical change observed for NaNO2. When hemoglobin changes to methemoglobin, Fe2+ changes to Fe3+, leading to hyperintensity on T1WI [82, 83].

Melanin is a metal-binding protein that binds various metals in vivo. The signal intensity of melanin is related to the binding of melanin to paramagnetic metals such as Cu2+, Fe3+, Zn2+, and Mn2+; however, it shows a particularly strong affinity for Fe3+ which affects the signal intensity of melanin on T1WI [84, 85]. Melanin appears in the brain as neuromelanin [86] and in the presence of metastatic melanoma, primary diffuse meningeal melanomatosis, neurocutaneous melanosis, and Aspergillus infection [8]. The location and extent of a region of high signal intensity on T1WI are useful diagnostic findings. Neuromelanin is present mainly in the substantia nigra and locus coeruleus within the pons, and shows high signal intensity on T1WI in normal human brain. A loss of high signal intensity in these areas on T1WI indicates a decrease in the amount of neuromelanin, which can lead to a diagnosis of Parkinson’s disease [86].

Copper

Copper is another essential trace element, i.e., an important cofactor for many enzymes. Copper is obtained orally from the daily diet and is excreted via the biliary system. Copper metabolic abnormalities include Wilson’s disease, which is a condition associated with excessive copper accumulation, and Menkes disease, which is a condition related to copper deficiency [87–90]. Wilson’s disease is caused by a mutation in the gene for ATP7B, which binds copper to ceruloplasmin and facilitates the excretion of the copper–ceruloplasmin complex in the bile; the mutation leads to the gradual accumulation of copper in the body [89–91]. It is known that copper has a relatively small paramagnetic effect on MRI [92, 93] and that, on its own, copper accumulation does not directly cause signal intensity changes [91]. Brain MRI abnormalities in Wilson’s disease patients include brain atrophy, symmetric regional hyperintensity on T2WI, hypointensity in basal ganglia on T2WI, and a hyperintense globus pallidus on T1WI [94–96]. The “face of giant panda” sign characterized by hypointensity in the red nucleus and substantia nigra surrounded by hyperintensity in the midbrain is known to be characteristic of patients with Wilson’s disease, although it is only rarely seen [96, 97]. Hyperintensity on T2WI is observed in putamen (72 %), caudate (61 %), thalami (58 %), midbrain (49 %), pons (20 %), cerebral white matter (25 %), cortex (9 %), medulla (12 %), and cerebellum (10 %) [96], but these signal changes are the opposite to those produced by the paramagnetic effect of copper. The cause of this hyperintensity on T2WI is unclear, but edema, gliosis, neuronal necrosis, and cystic degeneration due to copper toxicity have all been proposed. Hyperintense regions on T2WI occasionally diminish after therapy, and this change may reflect edema and gliosis [98–100]. Hyperintensity of the globus pallidus on T1WI was initially attributed to copper accumulation [101], but manganese deposition due to liver dysfunction has more recently been implicated [91, 100]. Hypointensity of the substantia nigra, striatum, and globus pallidus on T2WI is believed to be due to increased accumulation of both copper and iron (Fig. 3) [91, 96, 102]. Ceruloplasmin, which transports not only copper but also iron, is decreased in Wilson’s disease, and this enhances iron accumulation in the body [89, 90]. Indeed, in autopsy studies, both copper and iron accumulation were found to be higher in brains with Wilson’s disease than in control brains [103, 104]. In addition, a PET study with radioactively labeled iron in human subjects showed increased iron uptake by the brain in Wilson’s disease patients compared to healthy volunteers [105]. In Wilson’s disease, the paramagnetic effects of manganese and iron and the neurotoxicity of copper greatly affect the abnormality of brain MRI.

Fig. 3
figure 3

A 43-year-old female with Wilson’s disease. The globus pallidus (white arrows), putamen (blue arrows), and pulvinar of thalamus (yellow arrows) are hypointense compared with a normal age-matched female (Fig. 2b). The hypointensity is probably caused by iron deposition due to low ceruloplasmin

Full size image

Gadolinium

Gadolinium is used in MRI contrast media owing to the strong paramagnetic effect of the gadolinium ion. Because gadolinium is toxic to humans, it is injected along with chelating compounds that can prevent its toxicity by promoting its quick excretion from the body [106]. However, increased signal intensity in the dentate nucleus, globus pallidus, and pulvinar of thalamus on T1WI in patients subjected to repeated GBCA administration was recently reported for the first time. The administered gadolinium remains in the human body for a long time and causes abnormal signal intensity on T1WI (Fig. 4) [106–108]. Because of the toxicity of gadolinium, residual gadolinium has been attracting increasing attention [109–116]. These regions show increased signal with the administration of linear-type GBCAs but not macrocyclic-type GBCAs; because linear-type GBCAs are more unstable than macrocyclic-type GBCAs, it is strongly suspected that the change in signal is caused by the release of free gadolinium ion from the chelating compound in the GBCA [116–118]. In autopsy studies, gadolinium is detected in the brain tissue from patients with a history of GBCA administration using ICP-MS [119–121]. The signal intensity on T1WI and the residual gadolinium in these brain regions are both correlated with the total dose of GBCA administered, and the gadolinium deposition is clearly due to past GBCA administration [106]. Animal studies have also confirmed that the signal from the dentate nucleus on T1WI is affected by previous gadolinium administration, and that gadolinium is deposited [122, 123], but none of these clinical phenomena have been reported in connection with gadolinium deposition in the brain in the presence of normal renal function [106].

Fig. 4
figure 4

A 63-year-old man with a history of multiple linear-type GBCA administration. Hyperintense lesions are present on the globus pallidus (white arrow) and dentate nucleus (red arrow) on sagittal T1WI due to gadolinium deposition

Full size image

Mercury

Mercury occurs naturally in three major forms: elemental mercury, inorganic compounds, and organic compounds. Elemental mercury can cross the blood–brain barrier and cause neurotoxicity. The first outbreak of methylmercury poisoning occurred in Japan around Minamata Bay. The mercury was discharged by an industrial plant, and biological accumulation occurred in the fish and shellfish in this bay. The patients developed a progressive neurologic illness (Minamata disease) caused by the ingestion of methylmercury-contaminated seafood. In the acute phase, high signal intensity regions appear in brain white matter on T2WI [124, 125]. Organic mercury poisoning causes central nerve and peripheral nerve toxicity. In the chronic phase of organic mercury poisoning, atrophy of the calcarine area, cerebellum, and postcentral gyri is observed [125, 126]. However, no specific signal intensity due to histologically proven mercury deposition has been reported.