Evaluation of possible drug–drug interaction between gadoxetic acid and erythromycin as an inhibitor of organic anion transporting peptides (OATP)

Authors


Abstract

Purpose

To evaluate if erythromycin compromises liver-specific enhancement of gadoxetic acid; both compounds competing in organic anion transporting peptides (OATP) -mediated hepatocytic uptake.

Materials and Methods

The study was approved by institutional review board. Twelve healthy subjects (nine men, three woman; mean age, 38.7 years) were examined twice by MR imaging with prior administration of NaCl solution (placebo) or 1000 mg of erythromycin following a randomized sequence. Gadoxetic acid (0.025 mmol/kg body weight) was administered 15 min after the end of infusions. Pre- and 20 min postcontrast two-dimensional gradient-recalled-echo sequences were acquired. Relative enhancements of liver parenchyma and ratio of means were calculated from signal intensity measurements. Plasma levels of gadoxetic acid and erythromycin were determined and given in geometric means and coefficients of variation (CV).

Results

Concentration of erythromycin directly after end of infusion was 13.9 mg/L (CV 14.9%). Gadolinium plasma concentrations 5 min after gadoxetic acid administration were 138.7 μmol/L (CV 20.4%) after erythromycin infusion and 129.6 μmol/L (CV 22.8%) after placebo. Mean relative enhancements of liver parenchyma were 88.1 (SD 24.9%) after erythromycin infusion and 92.6 (SD 17.9%) after placebo. Ratio of relative enhancements was 0.951 (95% confidence interval, 0.833; 1.061; statistically not significant).

Conclusion

Coadministration of erythromycin has no effect on gadoxetic acid enhanced liver MR imaging. J. Magn. Reson. Imaging 2011;33:409–416. © 2011 Wiley-Liss, Inc.

GADOXETIC ACID (PRIMOVIST®, Bayer Schering Pharma AG, Germany) is an approved gadolinium (Gd) -based contrast agent, used in T1-weighted MR imaging of the liver to detect and characterize lesions with known or suspected focal liver disease (1–8). Hepatocytic uptake of gadoxetic acid and thus hepatic MR enhancement is mediated by the organic anionic transport peptide (OATP) family. A wide variety of structurally diverse endogenous compounds and xenobiotics also enter the hepatocytes by using the same system and thus may compete with the uptake of gadoxetic acid. In an animal study, Kato et al evaluated if the hepatic enhancement characteristics of gadoxetic acid were influenced by prior administration of different commonly used clinical pharmaceuticals (9). It was shown that particularly rifampicin administrated far above the clinical dose caused a decrease in the hepatic enhancement of gadoxetic acid. The reduced enhancement with coadministration of rifampicin was attributed to the competitive inhibition at the level of OATP-mediated gadoxetic acid uptake. Reasons for the absence of an effect for other compounds could be their different affinities to OATP, that they might be substrates of different subtypes of OATP, or that they might be substrates mainly of other transporters. At the present time, there are no clinical data available to describe the effects of OATP inhibitors on the hepatocytic uptake of gadoxetic acid in humans.

In this study, we used erythromycin as representative of drugs inhibiting OATP (Fig. 1), as it is likely to be used more commonly than rifampicin (10). Erythromycin is routinely prescribed for respiratory tract infections, severe enteritis, legionellosis, syphilis, acne, and gonorrhea. Erythromycin was infused intravenously to achieve a high systemic concentration during the moment of gadoxetic acid administration for liver MR imaging. The objective of this study was to evaluate if erythromycin compromises hepatocyte-specific enhancement after injection of gadoxetic acid.

Figure 1.

Transport proteins involved in the hepatic excretion of drugs. Transporters which are the focus of this study are marked in black. Hepatic excretion path of erythromycin marked in dotted lines, path of gadoxetic acid marked in double lines. MDR, multidrug resistance protein; MRP, multidrug resistance-associated protein; OATP, organic anion transporting polypeptide; OAT, organic anion transporter.

MATERIALS AND METHODS

Study Design

Twelve healthy subjects were enrolled in this single-center, open-label, randomized, cross-over study with blinded reading. Institutional review board approval for this prospective study was obtained. The protocol conformed to Good Clinical Practice guidelines, and all subjects gave written informed consent.

Each subject received two treatments with gadoxetic acid, one with prior infusion of 500 mL of isotonic NaCl solution as placebo, and one with prior administration of erythromycin dissolved in 500 mL of isotonic NaCl solution as active treatment. Six subjects were randomly assigned to sequence 1 (placebo, interval 7 days, active treatment), another 6 subjects to sequence 2 (active treatment, interval 7 days, placebo).

Study Volunteers

Twelve healthy volunteers (nine men and three woman) with a mean age of 38.7 years (age range, 27–48 years), with a mean weight of 78.7 kg (range, 64.3–99 kg), a mean height of 1.79 m (range, 1.66–1.98 m), and a mean body mass index of 24.4 kg/m2 (range, 21.2–27.3 kg/m2) were included in the study according to the following criteria: > 18 years of age, body mass index ≥ 20 and ≤ 30 kg/m2, not pregnant or nursing females using an effective contraception, signed informed consent.

Subjects were excluded in case of history of relevant diseases, especially incompletely cured pre-existing diseases, history of malignant tumors, known or suspected benign liver tumors, liver disorders; viral hepatitis, presence of hepatitis B antigen or hepatitis C or human immune deficiency virus antibodies, kidney diseases, glomerular filtration rate <30 mL/min/1.72 m2, contraindication to MR imaging or the use of Gd-containing contrast agents, history of severe allergic or anaphylactoid reaction to any allergen, regular intake of medication (except contraceptives), use of medication within 8 weeks before the first study drug administration, positive urine drug screening, clinically relevant deviations of the screened laboratory parameters from reference ranges (in especially enzymes reflecting liver function, bilirubin, ferritin), smoking, suspicion of drug or alcohol abuse, special diets, for example, strict vegetarian or low calorie diet.

The exclusion criteria were chosen to ensure that volunteers with specific risks for administration of the study drugs and/or volunteers with conditions which may have had an impact on the aims of the study were excluded. One single female patient revealed an incidentally detected focal liver lesion during the study MR imaging. The finding was classified as minor protocol deviation and the patient was leaved in all the evaluations.

OATP Inhibitor

Erythromycin (Amdipharm Ltd., Ireland) was administered in a dose of 1000 mg corresponding to the range of daily doses recommended for adults for most infections. The infusion solution was prepared using a vial containing 1492.5 mg of erythromycin lactobionate (= 1000 mg of erythromycin) and was administered over 60 min in 500 mL of isotonic NaCl solution.

Contrast Agent

Gadoxetic acid, belonging to the class of linear ionic Gd-based contrast agents, is a liver-specific hepatobiliary MR imaging contrast agent which is taken up selectively by hepatocytes by means of OATP. The biliary excretion is assumed to be mediated mainly by means of the multidrug resistance-associated protein (MRP2) (11, 12). The agent has a low binding to plasma proteins (10.7 ± 3.4 %) and a biodistribution study in human revealed a dose-independent renal (41.6–51.2% of dose) and biliary (43.1–53.2% of dose) elimination (8, 13). The contrast agent was administered in a dose 0.025 mmol/kg body weight corresponding to 0.1 mL/kg body weight. Gadoxetic acid exhibits an r1 relaxivity of 6.9 L mmol-1 s-1 at 1.5 Tesla (T) in plasma at 37°C (14).

Imaging Protocol

MR imaging was performed using a 1.5T MR system (Magnetom Avanto, Siemens Healthcare, Germany). The examination was covering the whole liver. Matrix coils (six-channel-spine plus six-channel-body coil) were used. Sequence used was T1-weighted two-dimensional gradient-recalled-echo with fat saturation (repetition time/echo time [TR/TE] 152/2.72, flip angle 70°) acquired in breath hold in two slaps of each 22 s. An effective matrix of 146 × 256 with a square field of view (FOV) as small as possible to include the liver with a slice thickness of 7 mm and an interslice gap of 0.7 mm was used.

Gadoxetic acid was administered 15 min after the end of erythromycin or placebo infusion (Fig. 2) using a bolus injection of the contrast agent by an automatic power injector (Spectris Solaris; Medrad, Pittsburgh, PA) with an injection speed of 1 mL/s followed by a saline chaser of 20 mL. The bolus injection was used to simulate the administration of gadoxetic acid in routine care for acquisition of both dynamic perfusion phases and hepatocyte-selective phases. In our study, the liver was acquired precontrast and 20 min postcontrast agent administration.

Figure 2.

Flow-chart of study design including time points for blood sampling, acquisition of pre- and postcontrast two-dimensional GRE images, and gadoxetic acid bolus injection.

Quantitative Analysis

The blinded image evaluation was performed in an image core laboratory by an independent radiologist experienced in liver imaging (10 years of experience in abdominal MR imaging). Subject and treatment information were masked, scalar information was preserved. Training was provided to the reader before the evaluation in form of a review of the appropriate study guidelines and a review of the image display software.

For signal intensity measurement, one region of interest (ROI) was placed in the liver parenchyma, avoiding areas of blood vessels and artifacts (Fig. 3). The ROI had a minimal diameter of 1 cm. ROI of the chosen postcontrast MR image was copied and pasted into the corresponding precontrast image to assure same location and size of ROI. All MR imaging sets (with prior erythromycin or placebo medication) provided to the reader in a randomized order.

Figure 3.

A 48-year-old male subject. Axial T1-weighted two-dimensional gradient-recalled echo sequences with fat saturation before (a,c), and 20 min after (b,d) administration of gadoxetic acid. Identical windowing and contrast setting (W1000/C500) of the four images allows for the direct comparison of the enhancement. The drawing of the liver parenchyma region-of-interest is depicted. The basic signal intensity between the two MRI examinations (a with prior administration of erythromycin; c with prior administration of placebo) acquired with an interval of 7 days attest the complete wash-out of gadoxetic acid. Relative enhancement of liver parenchyma with and without erythromycin medication (b,d) was highly comparable (80.9% and 85.2%, respectively).

The percentage of relative enhancement of the liver was based on signal intensity measurements obtained from the results of pre- and postcontrast MR imaging as follows:

equation image

with SIpost representing the signal intensity of the liver parenchyma postcontrast and SIpre representing the signal intensity of the liver parenchyma precontrast.

Determination of Gd and Erythromycin Concentration in Plasma

Overall, five blood samples were collected to determine gadoxetic acid and erythromycin (active treatment only) plasma concentrations. The sampling times in respect to the injection of gadoxetic acid are depicted in Figure 2.

Gd concentrations of the plasma samples were determined by validated inductive coupled plasma-mass spectrometric method. The lower limit of quantification for measuring Gd concentrations was 0.0636 μmol/L. The concentrations of erythromycin in plasma were determined using validated liquid chromatography method coupled with mass spectroscopy.

Population Pharmacokinetic Analysis of Gd Plasma Concentrations

All 96 Gd plasma concentrations (12 subjects, 2 treatment periods, 4 samples per period) were analyzed simultaneously using nonlinear mixed-effects modeling, version VI level 2.0 (Icon Development Solutions, Ellicott City, MD). Model development and qualification were conducted in agreement with the FDA guidance on population pharmacokinetic studies (15). The potential impact of treatment period and prior administration of erythromycin on the pharmacokinetics of gadoxetic acid was investigated as covariates. Significance was evaluated with a likelihood ratio test at a P value < 0.001.

Based on the model, the pharmacokinetic parameters total body clearance (CL), volume of distribution (V), area under the curve (AUC), plasma half-lives (t½ α, t½ β ), and maximum plasma concentration (Cmax) were determined. Individual values of V at steady-state (Vss) were calculated by summing up the individual post hoc estimates of the central V (V1) and the peripheral V (V2). Individual AUC values were derived by dividing the individual administered dose by the individual post hoc estimate of CL. Both CL and Vss were normalized to the individual body weight of the subject.

Statistical Analysis

All data were listed and all variables were summarized according to their type. Variables measured on continuous scales were summarized by use of descriptive statistics (number of volunteers, geometric mean, coefficient of variation (CV), median, minimum, and maximum). Variables measured on ordinal or nominal scales were summarized by use of frequency tables showing the number and percentage of volunteers falling within a particular category.

The primary analysis was the ratio of means of relative enhancements of signal intensity:

Mean relative enhancement after injection of gadoxetic acid with prior administration of erythromycin / Mean relative enhancement after injection of gadoxetic acid with prior administration of placebo.

For this ratio, a point estimate was given and the corresponding 95% confidence interval was calculated by using Fieller's theorem (16).

As a secondary analysis, the difference between the means of relative enhancement with prior administration of erythromycin and prior administration of placebo was analyzed. A point estimate for this difference was given, as well as its 95% confidence interval (based on the assumption that the primary variable follows a normal distribution). All statistical analyses were considered as purely explorative, no confirmatory analyses were intended.

The statistical evaluation was performed by using SAS version 9.1.3 (SAS Institute Inc., Cary, NC), MATLAB, version R2009a (The MathWorks Inc., Natick, MA), and S PLUS (S PLUS 6.2 for Windows XP SP2 and S-PLUS 6.2 for Linux, S-PLUS, Insightful Inc., Seattle, WA).

RESULTS

Plasma Concentration of Gadoxetic Acid and Erythromycin

The geometric mean plasma concentration of gadoxetic acid in the placebo group (coadministration of NaCl) and in the treatment group (coadministration of erythromycin) at 5 min, 25 min, 180 min, and 360 min after bolus injection are depicted in Figure 4. In the treatment group, geometric mean plasma concentrations of erythromycin 15 min before gadoxetic acid injection and at 5 min, 25 min, 180 min, and 360 min after gadoxetic acid injection were 13.9 mg/L (CV 14.9%), 7.98 mg/L (CV 15.9%), 6.80 mg/L (CV 18.4%), 2.97 mg/L (CV 23.7%), and 1.14 mg/L (CV 27.4%), respectively.

Figure 4.

Geometric mean and coefficient of variation of gadoxetic acid plasma concentration time profiles after administration of erythromycin (X) or placebo infusion (O). Slightly higher mean plasma gadolinium concentrations were observed at all time points with prior erythromycin administration, but these differences were very small in relation to the geometric SDs.

Enhancement

The mean relative enhancement of liver parenchyma after injection of gadoxetic acid (Figs. 3, 5) was 88.1 ± 24.9% after infusion of erythromycin (treatment group) and 92.6 ± 17.9% in the placebo group. The ratio of mean relative enhancements was 0.951 (95% confidence interval: 0.833 to 1.061). The difference of the mean relative enhancement was 4.5% in favor of the treatment sequence with prior placebo administration. Because the confidence interval for the ratio covers 1.0, no statistically significant treatment effect was observed.

Figure 5.

Boxplots for relative enhancement of signal intensity by treatment: Lines in the boxplots represent the median, whiskers extend to minimum and maximum measurements. Arithmetic means displayed by character “+”. Upper and lower borders of boxes represent 25th and 75th percentiles. The arithmetic means were different by 4.5% in favor of the treatment sequence with prior placebo administration.

Population Pharmacokinetic Analysis of Gd Plasma Concentrations

Gadoxetic acid pharmacokinetics could be adequately described by a linear open two-compartment model with elimination from the central compartment. Categorical covariate analysis for treatment period and prior administration of erythromycin showed no statistically significant influence on gadoxetic acid pharmacokinetics. Median values and the 2.5th and 97.5th percentiles of the value distributions of the main pharmacokinetic parameters are presented in Table 1.

Table 1. Individual Estimated and Derived Pharmacokinetic Parameters of Gadoxetic Acid After Single Intravenous Injection of 0.025 mmol Gd/kg Body Weight*
QuantityUnitsMedian estimate2.5th percentile97.5th percentile
  • *

    AUC, area under the plasma concentration versus time curve; CL/WGHT, total body clearance normalized to body weight; Cmax, maximum drug concentration in plasma after single dose administration; t1/2 α, half-life associated with the first slope in a biphasic decline; t1/2 β, half-life associated with the second (=terminal) slope in a biphasic decline, Vss/WGHT, volume of distribution at steady-state normalized to body weight.

CL/WGHTL/h0.1840.1530.215
Vss/WGHTL0.2920.2550.329
AUCμmol*h/L135.390114.778156.002
t1/2 αh0.2250.1900.259
t1/2 βh1.4161.2891.544
Cmaxμmol/L166.474164.110168.839

DISCUSSION

Traditionally, inhibition of major drug metabolizing enzymes (eg, cytochrome P450) leading to increased plasma concentrations of simultaneously administered compounds has been the focus of drug–drug interaction studies. Newly recognized, additional determinants of drug disposition are uptake transporters of the organic anion transporting polypeptides (rodents: Oatps, human: OATPs) family, and efflux (out of cell) transporters, which are located, for example, at the apical membrane of hepatocytes.

Members of the OATP group transport a wide range of drugs including HMG-CoA reductase inhibitors, benzylpenicillin, macrolid antibiotics, digoxin, fexofenadine hydrochloride, methotrexate, and rifampicin across the basolateral membrane of hepatocytes as the first step in hepatic elimination (10, 17–19). OATPs mediate the uptake of large amphiphatic organic anions, organic cations, and uncharged substrates and they exhibit broad substrate specificity with a degree of overlap, suggesting the possibility of transporter-mediated drug–drug interactions with other substrates. Clinical relevant OATP-mediated drug–drug interactions with OATP substrates in vivo triggered, for example, by cyclosporin A and rifampicin have been reported (20–22).

For the MR imaging contrast agent gadoxetic acid, interactions with rifampicin, prednisolone, doxorubicin hydrochloride, cisplatin, and propranolol hydrochloride have been observed in an animal in vivo study at very high substrate concentrations (9). A pharmacokinetic study of gadoxetic acid in healthy volunteers has shown that gadoxetic acid is not metabolized (13) and a metabolic interaction is, therefore, highly unlikely. Based on the preclinical data, an interaction on the transporter level was assumed. Therefore, the present study was performed to specifically investigate this aspect. In humans, OATP1B1 and OATP1B3 are presumably the most important transporters for gadoxetic acid uptake into hepatocytes (23, 24). The nomenclature of the OATP family was updated after the appearance of different names for the same transporter protein (25, 26) in the literature. OATPs/Oatps within the same family share ≥40% amino acid sequence identity and are designated by Arabic numbering (eg, OATP1). Individual subfamilies include OATPs/Oatps with amino acid sequence identities ≥60% and are designated by letters (eg, OATP1B). Individual gene products (proteins) within the same subfamily are designated by additional Arabic numbering (eg, OATP1B1). OATP1 is further differentiated into OATP1B1 (previous protein names LST-1, OATP-C, OATP2) (27) and OATP1B3 (previous protein names LST-2, OATP8) (25, 28, 29). OATPs are encoded by genes of the solute carrier organic anion transporters (SLCO) family (previously SLC21) and are located in the short arm of chromosome 12 (26). The estimation of the role of a single transporter in drug–drug interaction may be challenging, because, for example, many OATP1B1 substrates are also substrates of other drug transporters.

In vitro, van Montfoort et al observed a 51% inhibition of gadoxetic acid uptake in the presence of an unbound plasma concentration of 100 μM rifampicin (23) using Xenopus leavis oocytes after microinjection of rodent Oatp1. It has to be emphasized that this drug concentration was far above the therapeutic concentrations ranging between 0.2 and 3 μM in men (30). Comparable investigations using human OATP1 were, to the best of our knowledge, not repeated after the identification of the human analogues in the years 1999–2000 (27, 28). To evaluate in vivo the effects of OATP inhibitors at clinical doses on the hepatocytic uptake of gadoxetic acid, erythromycin was chosen for our study (26, 31). Macrolides including azithromycin, clarithromycin and erythromycin are frequently prescribed drugs for pneumonic infections as reported in a Canadian pneumococcal surveillance study. The annual macrolide prescription rate increased from 107 to 123 prescriptions/1000 persons per year between 1995 and 2005 (32). Erythromycin itself revealed a decreasing frequency of prescription as shown representatively in Europe and Canada while the frequency of newer macrolides azithromycin and clarithromycin increased significantly (32–34). Further indications for erythromycin are severe enteritis, mycoplasma and legionellosis, syphilis, acne, and gonorrhea. Because of its frequent use and its wide spectrum of indications, the likelihood of drug–drug interaction between macrolides and gadoxetic acid in routine care appeared higher than between rifampicin and the liver-specific MR imaging contrast agent. Furthermore, a consideration in the context of our study performed in healthy volunteers was erythromycin's more favorable adverse reaction profile.

An intravenous infusion was chosen for administration of erythromycin to achieve a high systemic concentration during the period of gadoxetic acid administration. Furthermore, the inter-individual variability in erythromycin concentrations were reduced due to the absence of the first pass effect observed after oral dosing. Erythromycin was administered in a dose of 1000 mg corresponding to the range of daily doses recommended for adults for most infections. In a study with a very similar erythromycin infusion (1000 mg diluted in 100 mL of isotonic saline solution, infusion with a constant flow over 1 hours, 9 female patients with a mean weight of 59.4 kg) Segui et al reported maximum serum concentrations 5, 20, 30, and 300 min after the end of infusion, of 11.6 mg/L, 7.5 mg/L, 6.8 mg/L, and 2.7 mg/L, respectively (35). We measured a plasma concentration of 13.9 mg/L directly after the end of infusion and 7.98 mg/L during the MR examination 5 min after gadoxetic acid administration (and respectively 20 min after end of erythromycin infusion). In our study, even 5 min after gadoxetic acid administration, the concentration in all subjects was higher than the maximum serum level of 6 mg/L of erythromycin after single oral administration of 1000 mg of erythromycin to healthy subjects (36). The plasma concentrations of erythromycin were expected to correlate with respective changes in the concentration in the extracellular space of the liver where competition of erythromycin with study drug at the transporter level may occur. Therefore, a clinically relevant concentration of erythromycin at the transporter was achieved during the gadoxetic acid administration thereby maximizing the likelihood to detect any potential interaction with the OATP-mediated liver uptake. Our results revealed no statistically significant differences in gadoxetic acid relative hepatic enhancement and only minor differences in the geometric mean plasma gadolinium concentration between the treatment sequences with prior placebo or with prior erythromycin administration. The slightly lower enhancement in the erythromycin sequence was consistent with the slightly higher mean gadolinium plasma concentrations: lower enhancement might be the result of decreased hepatic uptake, and might, therefore, result in slightly higher plasma concentrations. Furthermore, the result of similar enhancement between the treatments sequences excluded hypothesis of eventual increased enhancement due to longer retention of gadoxetic acid in the hepatocyte because of competition with a coadministered drug for the biliary excretion by the major canalicular transporters as reported by Kato et al in vivo for prednisolone, doxorubicin hydrochloride, cisplatin, and propranolol hydrochloride (9). We showed that simultaneous medication with erythromycin in patients with normal liver function and no pathological increase of bilirubin level should not interfere with liver-specific MR imaging using gadoxetic acid.

The prediction of the impact of other coadministered drugs classified as inhibitors of OATP on gadoxetic acid pharmacokinetics might be challenging because of the presence of different subtypes of OATP with different substrate affinities and specificities. Most substrates are not specific for a single OATP, and even in the situation of complete inhibition of one specific OATP, alternative uptake transporters are able to compensate this blockade. Second, half-maximum inhibitory concentration (IC50) of many OATPs were determined with very heterogeneous methods, as seen partially in large variance of the results: for example, erythromycin is known to inhibit both OATP1B1 (binding affinity of the inhibitor (Ki) of 11.4 μM, and IC50 217 μM) and OATP1B3 (IC50 34 μM) (10, 37). The respective Ki's of rifampicin are between 0.48 and 17 μM for OATP1B1; and 5 μM for OATP1B3 (10, 26, 37). These results determined in vitro demonstrate that both drugs are potent OATP inhibitors. However, it is difficult to predict the in vivo relevance of OATP inhibition based on these in vitro data as predictions involving transporters are still less well characterized as compared to, for example, metabolic drug–drug interactions.

A limiting factor of our study was the evaluation on the basis of the surrogate marker signal intensity measurement. The clinical endpoint would be the accuracy to detect and characterize focal liver lesions, which cannot be determined in healthy volunteers but would necessitate the inclusion of patients with hepatic pathologies.

In conclusion, no dose adjustment is necessary for the coadministration of gadoxetic acid and erythromycin in the general population. The possible interaction of gadoxetic acid and erythromycin in patients with decreased hepatic function and OATP capacity such as patients with liver cirrhosis should be assessed in further studies. These studies are relevant as gadoxetic acid-enhanced MR imaging has been recommended for characterizing nodules in liver cirrhosis and the liver-specific enhancement might be decreased in patients with advanced impaired liver function.

Acknowledgements

Hanns-Joachim Weinmann, whose name is closely related to the development of gadoxetic acid, for his uncomplaining and substantiated teaching about MR imaging contrast agents characteristics over the last 20 years. Hanns-Joachim Weinmann died in December 2009. We miss his personality and his passion for magnetic resonance imaging.

Ancillary