Evaluation of Gadolinium Deposition in Cortical Bone Using Three-Dimensional Ultrashort Echo Time Quantitative Susceptibility Mapping (3D UTE-QSM): A Preliminary Study

Abstract

Purpose:

We aimed to investigate the feasibility of three-dimensional ultrashort echo time Cones-based quantitative susceptibility mapping (3D UTE-Cones-QSM) for assessment of gadolinium (Gd) deposition in cortical bone.

Methods:

Forty tibial bovine cortical bone specimens were divided into five groups and were soaked in phosphate-buffered saline (PBS) solutions with five different Gd concentrations of 0, 0.4, 0.8, 1.2, and 1.6 mmol/L, respectively, for 48 hours. Eight rabbits were randomly allocated into three groups, including a normal-dose macrocyclic gadolinium-based contrast agent (GBCA) group (n=3), a high-dose macrocyclic GBCA group (n=3), and a control group (n=2). All bovine and rabbit tibial bone samples underwent magnetic resonance imaging (MRI) on a 3T clinical MR system. A 3D UTE-Cones sequence was utilized to acquire images with five different echo times (i.e., 0.032, 0.2, 0.4, 0.8, and 1.2 ms). The UTE-Cones images were subsequently processed with the morphology-enabled dipole inversion algorithm to yield a susceptibility map. The average susceptibility was calculated in three regions of interest in the middle of each specimen, and Pearson’s correlation between the estimated susceptibility and Gd concentration was calculated.

Results:

Bone samples soaked in PBS with higher Gd concentrations exhibited elevated susceptibility values. A mean susceptibility value of −2.47±0.23 ppm was observed for bovine bone soaked in regular PBS, while the mean QSM (quantitative susceptibility mapping) value increased to −1.75±0.24 ppm for bone soaked in PBS with the highest Gd concentration of 1.6 mmol/L. A strong positive correlation was observed between Gd concentrations and QSM values. The mean susceptibility values of rabbit tibial specimens in the control group, normal-dose GBCA group, and high-dose GBCA group were −4.11±1.52 ppm, −3.85±1.33 ppm and −3.39±1.35 ppm, respectively.

Conclusion:

A significant linear correlation between Gd in cortical bone and QSM value was observed. The preliminary results suggest that 3D UTE-Cones-QSM may provide sensitive non-invasive assessment of Gd deposition in cortical bone.

1. Introduction

Gadolinium-based contrast agent (GBCA) is widely used in magnetic resonance imaging (MRI) for clinical diagnosis of disease as well as research. More than 300 million doses have been used in clinics since the first Food and Drug Administration (FDA) approval of GBCA in 1987 [1]. Currently, about 30 million doses are administered in clinical MRI exams every year.

To date, there are nine different kinds of FDA-approved GBCAs available for clinical use. Based on their biochemical properties, the GBCAs can be classified as linear or macrocyclic, and nonionic or ionic, such as Gadoversetamide (liner nonionic), Gadopentetate dimeglumine (linear ionic), and Gadobutrol (macrocyclic nonionic) [2].

Macrocyclic agents have been shown to be more stable and less prone to release gadolinium (Gd) ions than their linear GBCA counterparts in in vitro experiments [4] and, as a result, are recommended by the National Institutes of Health as the preferential option in clinical practice [2]. At the same time, the European Medicines Agency has issued recommendations for the restriction of linear GBCAs in body scans. However, an increasing number of investigations have shown that even in patients with normal renal function, GBCAs at large—including those classified as macrocyclic—are not fully excreted from the body and that Gd has in fact been retained in the brain and other organs such as the skin and bone [3].

While Gd accumulation and retention in brain tissues are the main focus of research [5,6,7], a recent study has identified bone as a major storage site of Gd with up to 23 times higher concentrations than those found in the brain following Gd administration [3]. In fact, a study from early 2004 found that various GBCAs deposited differently in bone. Notably, about 1–2% of the total injection dose of linear contrast medium was found to deposit in bone for a long time [8]. In 2006, White et al. [9] used inductively coupled plasma mass spectroscopy (ICP-MS) to uncover Gd retention in bone specimens from patients with normal renal function who had undergone hip replacement surgery. In 2009, Darrah et al. [10] used ICP-MS to further demonstrate that there were anomalously high Gd concentrations in the bone tissues of patients who had been exposed to Gd-chelate contrast agents. Their findings suggested that Gd retention in bone tissues might be longer than 8 years. In 2016 the term “gadolinium deposition disease” was first proposed to describe “symptomatic deposition of gadolinium in individuals with normal renal function” by Semelka and his colleagues in the USA. In some patients, bone/joint pain and severe joint contractures of the limbs and neck were reported following exposure to gadolinium-based contrast agents [11,12]. A non-invasive technique to measure the Gd level in bone would be useful for monitoring Gd deposition [13].

Previously, Lord et al. was able to measure Gd retention in the tibia in vivo using an x-ray fluorescence system [14], but a non-ionizing radiation technique for the assessment Gd deposition is still lacking. Recently, quantitative susceptibility mapping (QSM) techniques have emerged as a promising MRI technique to estimate Gd intake [15,16]. Unfortunately, the conventional QSM technique cannot be used to assess Gd in bone due to the tissue’s extremely short T2* decay (< 1ms) [17]. Ultrashort echo time (UTE) MRI-based QSM has recently been investigated in the literature as a potential method to facilitate QSM in short T2 tissues [18–22].

In this work, we aimed to investigate the feasibility of assessing Gd deposition using a novel three-dimensional UTE-Cones based quantitative susceptibility mapping (3D UTE-Cones-QSM) technique. Bovine cortical bone samples were soaked in solutions with different concentrations of Gd, and rabbits were injected with different doses of macrocyclic GBCA. The magnetic susceptibility of each bone sample was assessed using the UTE-Cones-QSM technique at 3T.

2. Materials and Methods

2.1. Bovine Sample Preparation (in vitro)

40 tibial bovine cortical bone specimens harvested from freshly frozen bovine tibial midshafts were cut to 20 mm in length roughly at the midpoint of the tibial shaft using a Delta ShopMaster band saw (Delta Machinery, Tennessee, USA) [23]. The sectioned bone samples were divided into five groups, with eight samples per group, and were soaked in phosphate-buffered saline (PBS) solutions with five different Gd concentrations (Gadobutrol, Bayer Pharma AG, German): 0 (as a control group), 0.4, 0.8, 1.2, and 1.6 mmol/L for 48 hours. All bone samples were then embedded in 1% weight/volume agarose gel in two cylindrical plastic containers (10 cm in diameter, 20 samples in one container), where 20 bovine bones were embedded in each container.

2.2. Animals and Sample Preparation (in/ex vivo)

This prospective study was approved by the local Institutional Animal Care and Use Committee. Eight male New Zealand Rabbits (3–3.5 kg) were allowed one week to acclimatize under standard laboratory conditions at 22 °C with 12-hour light/dark cycles. Standard food and water were serviced ad libitum. Rabbits were then randomly allocated into three groups: a normal-dose macrocyclic GBCA group (n=3), a high-dose macrocyclic GBCA group (n=3), and a control group (n=2). The experimental design is consistent with our previous experiment [24]. Firstly, GBCAs were intravenously administrated for five consecutive days per week over a period of four weeks. The normal-dose macrocyclic GBCA group was injected with a daily dose of 0.3 mmol/kg bodyweight gadobutrol (Gd-DO3A-butrol, Gadovist; Bayer Vital GmbH), and the high-dose macrocyclic GBCA group was injected with 0.9 mmol/kg bodyweight gadobutrol. The control group was injected with a daily dose of 0.9 ml/kg bodyweight saline. Subsequently, rabbits were allowed a 4-week recovery, followed by sacrifice. The left tibia was harvested from each animal for dissection.

After clearing external muscle, other soft tissues, and bone marrow, each rabbit tibia was cut into approximately 3 cm segments and stored at −20°C. At the time of experimentation, the tibial samples were thawed in PBS solution at 4°C for 24 hours, then embedded in 1% weight/volume agarose gel in a 10-cm diameter cylindrical plastic container for MR imaging.

2.3. MR Imaging

QSM data were acquired on a 3T clinical MR system (MR 750, GE Healthcare, Milwaukee, WI, USA) using a previously reported 3D UTE-Cones sequence [37–39] which co-developed by our institute and GE Healthcare with a single channel transmit/receive birdcage wrist coil (Mayo Clinic, Rochester, MN, USA). Figure 1A shows the pulse sequence diagram for UTE-Cones imaging where TE was adjusted by delaying the readout gradients. The minimum TE was 0.032 ms due to the radiofrequency (RF) coil transmit/receive mode switching time in which data cannot be acquired. Figure 1B shows the 3D Cones k-space trajectory which allows efficient encoding of a 3D k-space [22,40,41].

Figure 1.

the pulse sequence diagram for ultrashort echo time cones imaging and the 3D Cones k-space trajectory.

The following imaging parameters were used for UTE-Cones imaging: flip angle = 15°; repetition time (TR) = 15 ms; TE = 0.032, 0.2, 0.4, 0.8, and 1.2 ms; readout bandwidth = 125 kHz, field of view = 60×60×50 mm³; acquisition matrix = 160×160×100; total scan time = 45 min (i.e., 9 min per scan; five scans to achieve five TEs). Note that images were acquired at five different TEs to achieve information regarding the phase evolution in free induction decay, which is the essential data for QSM processing.

2.4. Data Processing

All MR images were reconstructed using offline reconstruction code implemented in Matlab 2017b (MathWorks, Natick, MA, USA) based on the Nonlinear Fast Fourier Transform algorithm [25], where both magnitude and phase images were generated for each TE. The five complex MR images acquired at five different TEs were input to the subsequent QSM pipeline. First, a total field map was obtained using complex signal-based phase fitting and 3D region growing-based phase unwrapping [26]. Then, the background field was removed using the projection onto dipole field algorithm [27], which yielded a local field map (i.e., tissue field map). Finally, the morphology-enabled dipole inversion (MEDI) algorithm [28] was applied (with the Lagrange regularization parameter set to 5000 for the MEDI algorithm) to obtain the 3D susceptibility map [21]. The average susceptibility values were calculated by drawing 40 regions of interest (ROIs) for all 40 bone pieces. Each ROI was drawn through 6 adjacent slices in the middle of each bovine tibial specimen. The size of ROI ranged from ~300 to ~500 voxels. For the rabbit tibial specimens, one ROI was drawn for each bone piece, where the size of ROI ranged from ~60 to ~90 voxels.. The ROIs were manually drawn by an experienced radiologist (20-year experience).

2.5. Statistical Analysis

Statistical analysis was performed using SPSS (SPSS 24, IBM Corporation). The differences of susceptibility values across bovine tibial specimens with different Gd concentrations were analyzed using one-way analysis of variance with least significant difference post hoc. For all statistical analyses, P<0.05 was considered significant. The Pearson’s correlation between the measured susceptibility value of the cortical bone and Gd concentration was investigated using a linear regression analysis.

3. Results

Input images and the resultant susceptibility map from 3D UTE-Cones-QSM of the 40 bovine bone samples soaked in PBS with different Gd concentrations for 48 hours are shown in Figure 2. Input images and a representative susceptibility map from UTE-QSM in rabbit bone samples are shown in Figure 3.

Figure 2.

Three-dimensional ultrashort echo time Cones-based quantitative susceptibility mapping of 20 bovine bone samples soaked in five different groups of phosphate-buffered saline solutions with gadolinium concentrations. (A) The magnitude of input images, (B) the phase of input images, (C) the local field map from projection onto dipole field, and (D) the resultant susceptibility map in the axial and sagittal plane.

Figure 3.

Ultrashort echo time Cones-based quantitative susceptibility mapping in rabbit bone samples. (A) The magnitude of input images, (B) the phase of input images, (C) the local field map from projection onto dipole field, and (D) the resultant susceptibility map in the axial plane.

Tables 1 and 2 summarizes the susceptibility values of the bovine bone specimens soaked in PBS, with higher Gd concentrations exhibiting elevated susceptibility (i.e., less diamagnetic). A negative susceptibility value of −2.47±0.23 ppm was observed for the control group of bovine bones, while an elevated susceptibility value of −1.75±0.24 ppm was observed for the experimental group of bovine bones soaked in PBS with the highest Gd concentration of 1.6 mmol/L. There was a significant difference in susceptibility values between the control and experimental groups (P<0.05), as shown in Figure 4A. The linear regression analysis of the mean susceptibility values of bovine bone specimens versus Gd concentrations demonstrates a high linear correlation with a line equation: susceptibility = 0.48 [Gd] - 2.47 (ppm) (R² = 0.9839, P<0.05), as shown in Figure 4B.

Table 1.

The susceptibility values of all bovine bone specimens.

Gadolinium(mmol/l)	susceptibility value (ppm) (Mean±SD)
0	−2.47±0.23
0.4	−2.30±0.28
0.8	−2.07±0.50
1.2	−1.84±0.35
1.6	−1.75±0.24

The mean and standard deviation susceptibility values for five groups of bovine bone samples soaked in phosphate-buffered saline solutions with five different concentrations of gadolinium ranging from 0 to 1.6 mmol/L.

Table 2.

The mean and standard deviations of the susceptibility values for all 40 bovine bone specimens.

Gadolinium (mmol/l)	Susceptibility value (ppm) (Mean±SD)
	ROI 1	ROI 2	ROI 3	ROI 4	ROI 5	ROI 6	ROI 7	ROI 8
0	−2.33±0.23	−2.51±0.28	−2.92±0.18	−2.66±0.15	−2.44±0.71	−2.37±0.64	−2.14±0.63	−2.39±0.86
0.4	−2.56±0.25	−2.31±0.19	−2.75±0.13	−2.57±0.11	−2.01±0.67	−2.13±0.46	−2.03±0.48	−2.33±0.97
0.8	−1.65±0.21	−1.85±0.35	−2.49±0.80	−2.11±0.23	−2.24±0.86	−1.29±0.43	−2.01±0.45	−2.90±0.20
1.2	−1.83±0.34	−1.78±0.24	−2.21±0.41	−1.90±0.26	−1.99±0.73	−2.29±0.82	−1.31±0.63	−1.41±0.93
1.6	−1.74±0.39	−1.53±0.50	−1.91±0.31	−2.25±0.36	−1.61±1.09	−1.53±0.71	−1.62±0.44	−1.84±0.51

The mean and standard deviation susceptibility values for all 40 bovine bone specimens soaked in phosphate-buffered saline solutions with five different concentrations of gadolinium ranging from 0 to 1.6 mmol/L.

Figure 4.

(A) The difference in susceptibility values between the bovine bone specimens groups with different concentrations of gadolinium (0, 0.4, 0.8, 1.2, and 1.6mmol/L) (*P<0.05). (B) The linear regression analysis of the measured mean susceptibility value vs. gadolinium concentration demonstrates a high linear correlation with a line equation: susceptibility = 0.48 [Gd] - 2.47 (ppm) (R² = 0.9839, P<0.05).

The mean susceptibility values of rabbit tibial specimens in the control group, normal-dose GBCA group, and high-dose GBCA group were −4.11±1.52 ppm, −3.85±1.33 ppm, and −3.39±1.35 ppm, respectively. Increased QSM values were observed in tibial specimens from rabbits subject to the higher-dose GBCA. The linear regression analysis of the mean susceptibility values of rabbit tibia specimens versus gadolinium intake demonstrated a high linear correlation with a line equation: susceptibility = 0.79[Gd] – 4.1(ppm) (R² = 0.805, P<0.05), as shown in Figure 5.

Figure 5.

The linear regression analysis of the measured mean susceptibility value of rabbit tibia specimens vs. gadolinium intake demonstrates a high linear correlation with a line equation: susceptibility = 0.79 [Gd] - 4.1 (ppm) (R² = 0.805, P<0.05).

4. Discussion

Our data showed a high linear correlation between Gd in cortical bone and susceptibility values. These preliminary results suggest that 3D UTE-Cones-QSM may be a novel non-invasive imaging technique with high sensitivity for assessing Gd deposition in cortical bone. Zhang J et al. found a significantly positive correlation between the magnetic susceptibility of the GP in patients in the GBCA group and the number of Gd-DOTA injections. Choi Y et al. and Hinoda T et al. also found that the magnetic susceptibility of the globus pallidus increased after serial administration of gadobutrol [32,33,36]. Their studies showed that gadolinium deposition patterns in the body are consistent with ours.

The evaluation of Gd deposition in cortical bone is challenging because cortical bone has a very short transverse relaxation time and remains “invisible” with conventional clinical MR sequences. Gd deposition is expected to shorten T1 relaxation and shift bone susceptibility in the positive direction, but conventional T1 mapping techniques are incompatible with cortical bone applications due to this lack of detectable signal [29,30]. UTE T1 mapping has been investigated as a potential method for the evaluation of Gd deposition in cortical bone [31,24] and QSM has emerged as non-invasive imaging technique to assess Gd deposition in the brain [32,33]. However, like in the case of conventional T1 mapping, standard GRE QSM is not applicable to cortical bone due to the lack of detectable signal and the consequent unreliable phase information impairing susceptibility mapping. UTE-based QSM resolves these challenges by providing high signal from cortical bone, and thus reliable detection of phase evolution over TEs, which can overcome cortical bone’s extremely short T2* and facilitate accurate susceptibility mapping.

Gd is a strong paramagnetic substance with a molar susceptibility of 325 ppm·L/mol. As a result, susceptibility values in cortical bone should, in theory, increase with higher Gd concentrations. In our study, a negative susceptibility value of −2.47±0.23 ppm was observed in the control group of bovine bone, whereas the susceptibility value increased up to −1.75±0.24 ppm in the experimental groups of bones soaked in PBS with higher Gd concentrations (up to 1.6 mmol/L). Our results also showed increased QSM values in rabbit tibial specimens subject to higher-dose GBCA. There was a linear relationship between Gd in cortical bone and QSM values. These results indicate that 3D UTE-Cones-QSM can quantitatively detect Gd deposited in cortical bone.

The 3D UTE-Cones-QSM technique has many advantages for the quantification of Gd in cortical bone. Absolute quantification of contrast agent based on the T1/T2 effect requires calibration, is susceptible to B1 inhomogeneity (or flip angle errors), and is only accurate when the change in T1/T2 relaxation rate is linearly proportional to the contrast agent. Furthermore, the susceptibility information extracted from this QSM postprocessing may provide more specific information derived from morphological T1-weighted images, which provides a quantitative and direct estimation of susceptibility sources [33]. QSM, on the other hand, can overcome both the R2* saturation problem and R2* blooming artifact [34]. The 3D UTE-Cones sequence is not only highly effective in imaging short T2 tissues such as cortical bone but greatly reduces the total scan time by using efficient spiral cones sampling as well as anisotropic field of view and spatial resolution.

In addition to its capability to detect Gd in cortical bone, the 3D UTE-Cones-QSM approach can also be used to quantitatively detect other substances in tissues with strong magnetic susceptibility and short transverse relaxation times (e.g., iron, calcium, etc.). For example, some studies have used the UTE-QSM technique to detect mineral density in human femur and tibia cortical bone specimens, and found that the estimated susceptibility and bone mineral density were significantly correlated [18,23]. It has also been demonstrated that 3D UTE-QSM can detect hemosiderin deposition (i.e., iron) in hemophilic arthropathy [22].

Our study had several limitations. First, 3D UTE-Cones-QSM was used to quantitatively evaluate Gd in only cortical bone in this study. Studies on Gd in trabecular bone are lacking given that trabecular bone imaging has always been a technical challenge, but previous research has shown that Gd absorption in trabecular bone is greater than that in cortical bone [10]. At present, some researchers [35] have proposed that 3D inversion recovery UTE-Cones with a short TR/inversion time combination can provide selective imaging and quantitative assessment for short T2 tissues such as trabecular bone. Follow-up studies are needed to investigate UTE-QSM of trabecular bone. Second, our experiment was done ex vivo, with soft tissues surrounding the bone removed, a situation which may differ from the physiological situation. Furthermore, our experiment studied Gd concentration in healthy bone tissues. It will be necessary to better understand the Gd deposition in the cortical bone of not only patients with conditions such as diabetes, osteoporosis, and renal osteodystrophy, but also in children or pregnant women, as these various states may affect biological distribution [2]. Third, the sample size in this study was small, particularly the investigation of rabbit cortical bone samples. It will be necessary to increase the sample size in animal experiments in the future for a more systematic study of the relationship between Gd deposition in cortical bone and QSM. Lastly, the spatial resolution of the current imaging protocol for bone QSM was limited to compensate for the low signal-to-noise ratio (SNR) caused by the extremely short T2* and low proton density of bone. The low resolution can cause Gibbs ringing, which can be a confounding factor for QSM. In our future studies with animals, we will use a surface coil that can yield much higher SNR and hence higher spatial resolution.

In conclusion, our study shows a significant linear correlation between Gd in cortical bone and QSM values. The 3D UTE-Cones-QSM technique may be a novel, non-invasive technique that can provide quantitative assessment of Gd deposition in cortical bone. Its potential remains to be verified in clinical practice in the future.

Abbreviations

3D UTE-Cones-QSM

three-dimensional ultrashort echo time Cones-based quantitative susceptibility mapping

FDA

Food and Drug Administration

GBCA

gadolinium-based contrast agent

Gd

gadolinium

ICP-MS

inductively coupled plasma mass spectroscopy

MEDI

morphology-enabled dipole inversion

MRI

magnetic resonance imaging

PBS

phosphate-buffered saline

QSM

quantitative susceptibility mapping

RF

radiofrequency

ROI

region of interest

SNR

signal-to-noise ratio

TR

repetition time

UTE

ultrashort echo time

Data availability statement

The datasets used or analyzed during the current study are available from the corresponding author on reasonable request.