• Open Access

MR imaging of human pancreatic cancer xenograft labeled with superparamagnetic iron oxide in nude mice

Authors


X. M. Zhang, Sichuan Key Laboratory of Medical Imaging, Department of Radiology, Affiliated Hospital of North Sichuan Medical College, Sichuan 637007, China. E-mail: cjr.zhxm@vip.163.com

Abstract

The aim of this work was to investigate the MRI findings on tumor xenografts induced in nude mice by the inoculation of human pancreatic cancer cells labeled with superparamagnetic iron oxide (SPIO), and to monitor the kinetics of SPIO distribution in tumor xenografts. The labeled cancer cells were subcutaneously inoculated into 11 nude mice to induce tumor xenograft. The unlabeled cancer cells served as a control inoculated into nine nude mice. MR imaging was performed with a 1.5 T MR scanner for the tumor xenograft at the first, second and third week after the inoculation. We found that the tumor xenograft was induced in 100% nude mice on MR imaging for both groups in the first week after the inoculation. In the SPIO group, the tumors showed homogeneous hypointensity on T1- and T2-weighted and FIESTA images 1 week after inoculation. Two and 3 weeks after inoculation, the center of the tumors was still hypointense on all the above sequences. The tumor periphery was isointense on T1-weighted, and hyperintense on T2-weighted and FIESTA images. The tumors in control group were homogeneously hypointense or isointense on T1-weighted, and hyperintense on T2-weighted and FIESTA images in the first, second and third week after the inoculation. The size and signal-to-noise ratio of the tumor center in the SPIO group had decreased subsequent to the inoculation in all T1- and T2-weighted images and FIESTA. Our results showed the human pancreatic cancer cells labeled with SPIO can induce tumor xenograft in nude mice and MRI can monitor the kinetics of SPIO distribution in tumor xenografts. Copyright © 2012 John Wiley & Sons, Ltd.

1 INTRODUCTION

Pancreatic cancer is a devastating and virtually unexceptionally lethal malignancy, and is commonly diagnosed at incurable stages. It is the fourth leading cause of death among all cancers in the USA, with a dismal 5-year survival rate of less than 5% [1]. Despite much research and clinical treatment efforts aimed at better understanding the etiology and underlying pathophysiology, the growth patterns of pancreatic cancer are still unclear [2-4].

Molecular imaging is defined as the characterization and measurement of biological processes at the cellular and molecular levels [5]. All the major imaging modalities are contributing to this new field. Magnetic resonance imaging (MRI) is particularly attractive because of its superior spatial resolution and the harmlessness of the magnetic field used. Combined with an advanced contrast agent, MRI allows for molecular profiling of target tissues/cells and potentially enables the early detection of malignant diseases, more accurate staging and treatment monitoring [6].

Superparamagnetic iron oxide nanoparticles (SPIO) are used as a contrast agent and have been widely applied in the magnetic resonance molecular imaging [6-12]. Montet et al. [13] performed imaging of pancreatic ductal adenocarcinoma (PDAC) using functional SPIO targeted to bombesin (BN) receptors present on normal acinar cells of the pancreas. The BN peptide–nanoparticle conjugate, BN-CLIO (Cy5.5), decreased the T2 of normal pancreas and enhanced the ability to visualize tumor in a model of pancreatic cancer by MRI. In another study, Kelly et al. [14] demonstrated that plectin-1 targeted peptides conjugated to SPIO enabled detection of small PDAC and precursor lesions in engineered mouse models. Recently, Yang et al. [15] also reported that the systemic delivery of urokinase plasminogen activator receptor (uPAR)-targeted SPIO led to their selective accumulation within tumors of orthotopically xenografted human pancreatic cancer in nude mice. However, to our knowledge, there is no report involving monitoring the kinetics of SPIO distribution in SPIO-labeled pancreatic cancer cells following the growth of tumor xenografts.

In this study, human pancreatic cancer cells were labeled with SPIO and then inoculated into nude mice. We hypothesized that: (1) the SPIO-labeled pancreatic cancer cell would induce tumor xenografts after inoculation; and (2) iron distribution in the tumor xenografts would be heterogeneous and would be depicted on MR imaging.

2 MATERIALS AND METHODS

2.1 Cell Culture and Xenografts

All studies involving animals were approved by the institute's animal care and use committee. Poly- l-lysine (PLL), 0.1% (MW 150 000; Sigma Chemical Co., St Louis, MO, USA) was mixed with SPIO or Resovist (SHU555A; Bayer Schering Pharma, Berlin, Germany) for 1 h at room temperature on a rotating shaker to obtain PLL-Resovist complexes [16]. The PLL-Resovist complexes were added into RPMI-1640 medium containing newborn calf serum and the iron concentration was adjusted to 42 µg ml−1. The pancreatic cancer cell (PANC-1; Sichuan University) suspension was added to flask with RPMI-1640 medium containing newborn calf serum followed by incubation at 37 °C in a humidified incubator containing 5% CO2–95% air. When the PANC-1 cell confluence reached 80%, SPIO containing RPMI-1640 medium were added followed by incubation at 37 °C in humidified air with 5% CO2 for 24 h.

The supernatant was removed and cells were washed with Phosphate Buffered Saline (PBS) three times. Following digestion with 0.25% trypsin, cells were re-suspended in PBS to achieve to 1 × 107 cells ml−1 PBS. In the SPIO group, SPIO-labeled PANC-1 cells (0.1 ml) were subcutaneously injected into nude mice once at left axilla (n = 11). In the control group, unlabeled PANC-1 cells (0.1 ml) were subcutaneously inoculated into nine nude mice once at the left axilla. The procedures were carried out under aseptic conditions and then mice were normally housed.

2.2 MR Imaging Techniques

MR imaging was performed on a 1.5 T MR scanner (GE Medical Systems, Milwaukee, WI, USA) with a dedicated animal radio frequency coil (5 cm diameter, GE Healthcare Company, China) at 1, 2 and 3 weeks after inoculation. The MR protocol included spin echo (SE) T1-weighted imaging, spoiled gradient echo (SPGR) T1-weighted imaging, fast recovery fast spin echo (FRFSE) T2-weighted imaging and three-dimensional (3D) fast imaging with steady-state acquisition (FIESTA). The parameters were as follows: TR 300.0 ms, TE 15.0 ms, slice thickness 2.0 mm, intersection gap 0.5 mm, field of view (FOV) 6 × 6 cm, reconstruction matrix 256 × 160 for SE T1-weighted imaging; TR 185.0 ms, TE 4.2 ms, slice thickness 2.0 mm, gap 0.5 mm, FOV 6 × 6 cm, matrix 256 × 192 for SPGR T1-weighted imaging; TR 2160.0 ms, TE 88.5 ms, slice thickness 2.0 mm, gap 0.5 mm, FOV 6 × 6 cm, matrix 256 × 160 for FRFSE T2-weighted imaging; and TR 8.3 ms, TE 2.0 ms, slice thickness 1.0 mm, gap −0.5 mm, FOV 6 × 6 cm, matrix 160 × 160 for 3D FIESTA.

The nude mice in both groups were anesthetized intraperitoneally with 0.03 ml of chloral hydrate. Mice were kept in a supine position and put into animal coil followed by performing MR scanning. After scanning axial, sagital and coronal scout, axial T1- and T2-weighted imaging, and 3D FIESTA was acquired covering the transplanted tumor.

2.3 Analysis of MR Images

The original MR imaging data was loaded onto a workstation (GE, AW4.1, Sun Microsystems, Palo Alto, CA, USA) to be reviewed. One observer (with 4 years’ experience in interpreting abdominal MRI examinations) reviewed MR images of the tumor xenograft.

The signal intensity pattern of the xenograft in nude mice was studied on the above MR sequences for the two groups. The tumor xenografts were measured at their largest and smallest diameters; these measurements were then multiplied to produce the diameter products. We refer to the diameter products as the ‘size’ of the tumor xenografts. In the SPIO group, the size of the hypointense tumor center primary observed was also measured at the first, second and thirst week after inoculation.

The signal intensity of tumor xenograft was measured on axial T1-weighted, T2-weighted and 3D FIESTA images for both groups at different times after inoculation. The size of the region of interest (ROI) was 5 × 5 mm2. Three sites at the center of three equal portions in the same slice were measured at the tumor xenograft. The measured signal intensity was normalized to the signal intensities of background in the same encoding direction, and was defined as the signal-to-noise ratio (SNR). The average SNR of the three sites was considered the result of the tumor xenograft.

2.4 Histological Examination

Following MR scanning, three mice randomly selected from each group were sacrificed at designed time points and the tumor was collected. The tumor tissues were embedded in 10% buffered formaldehyde, then in paraffin for hematoxylin and eosin (H&E) staining and Prussian blue staining, and microscopy subsequently performed. One pathologist with 5 years’ experience in tumor pathology reviewed microscopic images of the tumor.

2.5 Statistical Analysis

Results were expressed as means ± standard deviation for quantitative data. The size and SNR of the tumor xenograft in both groups at the same sequence were compared with the independent-samples Student's t-test. In the SPIO group, the SNR of the tumor xenograft center and periphery obtained by the same sequence were analyzed with the paired-samples Student's t-test. The size of the hypointense tumor xenograft center in the SPIO group at different times was compared with the Kruskal–Wallis test. We tested the relationship of the size and the SNR of the tumor xenograft to the inoculation time with Student's t-tests, and to the continuous variables with Pearson's product moment correlations.

All of the statistical analyses in this study were performed with Statistical Package for Social Sciences for Windows (version 13.0, Chicago, IL, USA). A value of p < 0.05 was considered statistically significant.

3 RESULTS

3.1 Cell Culture and Xenografts

Human pancreatic cancer cells labeled (n = 11) and unlabeled with SPIO (n = 9) were subcutaneously inoculated in nude mice. The tumor incidence was 100% (20/20) for both groups in the first week after the inoculation. In the second and third weeks after the inoculation, the tumors continued to grow in the two groups with similar growth curves (data not shown).

3.2 MRI Findings of the Tumor Xenograft

3.2.1 MRI findings 1 week after inoculation

In the SPIO group of 11 tumor-bearing mice, the majority of the tumor was hypointense on T1- and T2-weighted images and FIESTA. A minimal periphery of the tumor showed isointense on T1-weighted and hyperintense on T2-weighted images (Fig. 1), and it was difficult to measure its signal intensity. In the control group of nine tumor-bearing mice, the tumor homogeneously showed isointense on T1-weighted, and hyperintense on T2-weighted images and 3D-FIESTA (Fig. 2).

Figure 1.

Magnetic resonance imaging of tumor xenograft in the superparamagnetic iron oxide group 1 week after inoculation. A majority of the tumor was hypointense on T1- and T2-weighted and fast imaging with steady-state acquisition (FIESTA) images (arrows).

Figure 2.

Magnetic resonance imaging of tumor xenograft in control group 1 week after inoculation. The tumor was homogeneously isointense on T1-, and hyperintense on T2-weighted and fast imaging with steady-state acquisition (FIESTA) images (arrows).

The size of the tumor xenograft was, respectively, 0.47 ± 0.15, 0.41 ± 0.13 and 0.49 ± 0.10 cm2 for the whole tumor and the hypointense tumor center in the SPIO group, and the whole tumor in the control group. There were no significant differences in the tumor size among them (p = 0.744).

The SNR of the tumor center in SPIO group was significantly lower than in the control group on all T1- and T2-weighted images and FIESTA (p < 0.05; Table 1).

Table 1. Signal-to-noise ratio of tumor xenograft 1 week after inoculation
 T1-SET1-SPGRT2-FRFSE3D-FIESTA
  1. p-Values indicated the comparisons between superparamagnetic iron oxide (SPIO) and control groups.

  2. SE, spin echo; SPGR, T1-weighted imaging, spoiled gradient echo; FRFSE, T1-weighted imaging, fast recovery fast spin echo; FIESTA, fast imaging with steady-state acquisition.

SPIO (n = 11)3.47 ± 1.544.22 ± 2.031.90 ± 0.692.49 ± 1.53
Control (n = 9)6.94 ± 1.7412.36 ± 4.876.68 ± 2.5424.44 ± 16.78
p-Valuep = 0.000p = 0.000p = 0.000p = 0.000

3.2.2 MRI findings 2 weeks after inoculation

In the SPIO group of eight tumor-bearing mice, heterogeneous signal intensity was found on T1- and T2-weighted images and FIESTA. The tumor center was still hypointense on all the above sequences. The tumor periphery was isointense on T1- and hyperintense on T2-weighted images and FIESTA (Fig. 3). In the control group of six tumor-bearing mice, the tumor xenograft was homogeneously isointense on T1-weighted, and hyperintense on T2-weighted and FIESTA images (Fig. 4).

Figure 3.

Magnetic resonance imaging of tumor xenograft in the superparamagnetic iron oxide group 2 weeks after inoculation. The tumor periphery was isointense on T1-, and hyperintense on T2-weighted and 3D fast imaging with steady-state acquisition (FIESTA) images (solid arrow). The tumor center remained hypointense (virtual arrows).

Figure 4.

Magnetic resonance imaging of tumor xenograft in control group 2 weeks after inoculation. The signal intensity patterns of the tumors were similar to that in the first week.

The size of the tumors was, respectively, 1.19 ± 0.19 and 1.15 ± 0.34 cm2 in the SPIO and control groups (p = 0.832). They were larger than in the first week (p = 0.008 for the SPIO group, and p = 0.004 for the control group). However, the size of the hypointense tumor center in the SPIO group was significantly smaller than in the first week (0.25 ± 0.09 vs 0.41 ± 0.13 cm2, p = 0.008).

The SNR of the tumor center in the SPIO group was significantly lower than that of the tumor periphery in the SPIO group and that in the control group (p < 0.05, Table 2) on all of the above MR sequences. The SNR of the tumor peripheral in the SPIO group was also significantly lower than that in the control group on T1- and T2-weighted images, except on the T1 SPGR sequence (Table 2).

Table 2. Signal-to-noise ratio (SNR) of tumor xenograft 2 weeks after inoculation
 T1-SET1-SPGRT2-FRFSE3D-FIESTA
  1. a, Comparison between the SNR of the tumor center in the superparamagnetic iron oxide (SPIO) group and that in the control group. b, Comparison between the SNR of the tumor periphery in SPIO group and that in the control group. c, Comparison between the SNR of the tumor center and the periphery in the SPIO group.

  2. SE, spin echo; SPGR, T1-weighted imaging, spoiled gradient echo; FRFSE, T1-weighted imaging, fast recovery fast spin echo; FIESTA, fast imaging with steady-state acquisition.

Tumor center in SPIO group (n = 8)2.52 ± 1.283.72 ± 1.491.74 ± 0.492.24 ± 1.34
Tumor periphery in SPIO group (n = 8)6.13 ± 2.4318.12 ± 4.175.44 ± 1.3016.58 ± 4.56
Tumor in control group (n = 6)8.65 ± 2.6418.43 ± 4.636.70 ± 1.9613.75 ± 4.18
Independent-samples t-testa = 0.000a = 0.000a = 0.000a = 0.000
b = 0.000b = 0.689b = 0.000b = 0.000
Paired-samples t-testc = 0.000c = 0.000c = 0.000c = 0.000

3.2.3 MRI findings 3 weeks after inoculation

The signal intensity patterns of the tumor xenograft at the third week was similar to that at the second week after inoculation on T1- and T2-weighted images and FIESTA for both groups (Figs. 5 and 6).

Figure 5.

Magnetic resonance imaging of tumor xenograft in the superparamagnetic iron oxide group 3 weeks after inoculation. The tumor periphery was isointense on T1-, and hyperintense on T2-weighted and 3D fast imaging with steady-state acquisition (FIESTA) images (solid arrow). The tumor center remained hypointense, but was much smaller (virtual arrows).

Figure 6.

Magnetic resonance imaging of tumor xenograft in control group 3 weeks after inoculation. The tumor was larger than in the first and second weeks. However the signal intensity patterns of the tumors were similar to those seen previously.

The size of the tumor was 1.91 ± 0.56 cm2, which was similar to the size of 2.34 ± 1.24 cm2 in the control group (p = 0.515). The size of the tumor in both groups at the third week was larger than that at the second week (p = 0.041 for the SPIO group; p = 0.237 for the control group). However, the size of hypointense tumor center was significantly smaller than that at the first week after inoculation (0.19 ± 0.09 vs 0.41 ± 0.13 cm2, p = 0.004), and was slightly smaller than that at the second week after inoculation (0.19 ± 0.09 vs 0.25 ± 0.09 cm2, p = 0.278).

The SNR of the tumor center in the SPIO group was significantly lower than that of the tumor periphery in the SPIO group and that in control group on T1- and T2-weighted images and FIESTA. The SNR of the tumor periphery in the SPIO group was also significantly lower than that in control group on all the above sequences (p < 0.05, Table 3).

Table 3. Signal-to-noise ratio (SNR) of tumor xenograft 3 weeks after inoculation
 T1-SET1-SPGRT2-FRFSE3D-FIESTA
  1. a, Comparison between the SNR of tumor center in the superparamagnetic iron oxide (SPIO) group and that in the control group. b, Comparison between the SNR of the tumor periphery in SPIO group and that in the control group. c, Comparison between the SNR of the tumor center and the periphery in the SPIO group.

  2. SE, spin echo; SPGR, T1-weighted imaging, spoiled gradient echo; FRFSE, T1-weighted imaging, fast recovery fast spin echo; FIESTA, fast imaging with steady-state acquisition.

Tumor center in SPIO group (n = 5)1.88 ± 0.662.04 ± 1.181.28 ± 0.501.54 ± 0.62
Tumor periphery in SPIO group (n = 5)5.42 ± 1.2514.13 ± 1.474.63 ± 0.9716.10 ± 2.56
Tumor in control group (n = 4)8.92 ± 1.5419.47 ± 2.167.96 ± 1.6724.20 ± 10.94
Independent-samples t-testa = 0.000a = 0.000a = 0.000a = 0.000
b = 0.000b = 0.000b = 0.000b = 0.001
Paired-samples t-testc = 0.000c = 0.000c = 0.000c = 0.000

3.2.4 Correlation of the size and signal intensity to the duration after inoculation

The size of tumor both in the SPIO and control groups tended to increase subsequent to inoculation (r = 1.0 and 0.99, respectively; p = 0.10). However the size of the hypointense tumor center in the SPIO group tended to decrease following inoculation (r = −0.97, p = 0.16; Fig. 7).

Figure 7.

The correlation of the size of tumor xenograft to the time after inoculation. The size of tumor both in superparamagnetic iron oxide (SPIO) and control groups tends to increase following the time after inoculation. However the size of hypointense tumor center in the SPIO group tends to decrease with inoculation time.

The SNR of the tumor center in the SPIO group decreased subsequent to inoculation in all T1- and T2-weighted images and FIESTA (T1-SE, r = 0.99, p = 0.07; T1-SPGR, r = 0.96, p = 0.18; T2-FRFSE, r = 0.96, p = 0.17; FIESTA, r = 0.97, p = 0.17; Fig. 8). However, the SNR of tumor periphery in the SPIO group and that in the control group had no relationship to the time after inoculation.

Figure 8.

The correlation of the signal-to-noise ratio (SNR) of the tumor center in the superparamagnetic iron oxide group to the time after inoculation. The SNR decreased following the time after inoculation in all T1- and T2-weighted imaging and fast imaging with steady-state acquisition (FIESTA).

3.3 Pathological Features

H&E staining showed that tumor cells were different in size and morphology. The nucleus of the tumor cells was enlarged and deeply stained, and the nucleus–cytoplasm ratio was increased with an obvious karyosome. The morphology of the tumor cells in both the SPIO and control groups was similar in the first, second and third week after inoculation.

In the center of the tumor, an irregular necrotic area was found in the second and third weeks after inoculation. Karyopyknosis, nuclear fragmentations and unorganized materials were found in it. A bigger necrotic area in the tumor was found in the third week. Abundant blood capillaries were found in tumor periphery. Many brown granules were found in the center of tumor, while only a few brown granules were found in the periphery of tumor in the SPIO group. No brown granules were found in the control group (Fig. 9).

Figure 9.

Hematoxylin and eosin staining of tumors in the first week after inoculation (400×). In the superparamagnetic iron oxide (SPIO) group tumor cells were variable in size and morphology. Many brown granules were found at the center of tumor (A) (arrows), but not at the periphery of the tumor (B). In the control group, the morphology of the tumor cells is similar to that in the SPIO group; no brown granules were found in the center (C) or the periphery (D) of the tumor.

Prussian blue staining showed that blue-staining granules were regionally aggregated and mainly found in the center of the tumor xenograft. However, fewer blue-staining granules were noted in the tumor periphery. The further the distance from the center of tumor xenograft, the lower the iron content was (Fig. 10). The features of the tumor on Prussian blue staining in the second and third week were similar to that in the first week, but blue staining granules in tumor center was less than that in the first week.

Figure 10.

Prussian blue staining of tumor in superparamagnetic iron oxide group at the first week after inoculation (400×). Many blue granules (arrows) were found in the central area of the tumor (A); only a few blue granules (arrows) were found in the periphery of tumor (B); and at the edge of tumor (C), no blue granules were found.

4 DISCUSSION

Currently, using SPIO to label cells encompasses a balance between introducing a large quantity of SPIOs into the cell and the consequential cytotoxicity associated with the presence of SPIOs and the labeling techniques [8, 17-20]. Generally, the toxicity has a strong dependency on the physicochemical properties of SPIO, such as size, surface charge and surface coating materials, in addition to the dosage of SPIO and the duration of exposure. For example, the use of Feridex-labeled mesenchymal stem cells (MSCs) for chondrogenic MR tracking studies should be cautious because of the chondrogenic activity blocking of Feridex [17]. PLL is one of the most effective polycationic transfection agents studied to date and we have demonstrated that PLL–Resovist complexes produce sufficient and biocompatible cellular MRI contrasts for cell trafficking [16]. In the present study we found that human pancreatic cancer cells labeled with SPIO induced tumor xenograft in nude mice. The tumor incidence was 100% in both the SPIO and control groups. Our results indicated that our SPIO labeling system did not influence the activity of pancreatic cancer cells.

We previously labeled MSCs with SPIO and confirmed the blue granules in MSCs in Prussian blue staining as SPIO nanoparticles [16, 21]. We also found that the number of iron-containing cells decreased with the culture passage number after labeling in vitro[16]. In this study, we labeled human pancreatic cancer cell with SPIO, and found brown and blue granules in tumor xenografts in H&E staining and Prussian blue staining, respectively. The brown and blue granules should be iron oxide granules or SPIO. Our results showed that iron was mainly distributed in the center of the tumor xenograft, which caused hypointensity on T1- and T2-weighted and 3D FIESTA images.

The signal intensity of tumor xenograft in the SPIO group varied following the time of inoculation. In the first week, a majority of the tumor was hypointense on T1- and T2-weighted images and FIESTA, and in the second and third week after inoculation, the tumor center remained hypointense on all of the above sequences. The tumor periphery was isointense on T1-, and hyperintense on T2-weighted images and FIESTA. The signal intensity of the tumor in the control group was similar at 3 weeks. In Prussian blue staining, the features of the tumor on Prussian blue staining in the second and third week were similar to those in the first week, but blue granules in the tumor center in the last 2 weeks were less abundant than during the first week. Interestingly, the size of the hypointense tumor center had a decreasing tendency following the time of inoculation. The possible reason for this may be that cells in the tumor periphery proliferated more quickly than those of the tumor center, and the iron oxide contents decreased subsequent to the inoculation with cell division and degradation of SPIO [16, 21, 22].

Simon et al. [23] demonstrated higher T1 and T2 relaxivity values of free extracellular SPIO as opposed to intracellularly compartmentalized SPIO at 1.5 and 3 T. It seems that cellular compartmentalization of SPIO can change proton relaxivity [23]. In our study the SNR of the tumor center in the SPIO group decreased subsequent to inoculation in all T1− and T2-weighted images and FIESTA. In histology, in the center of the tumor, an irregular necrotic area and many brown granules were found in the second and third week after inoculation. The possible reason for the SNR patterns may be that the combination of extracellular with intracellular iron changed the proton relaxivity in the center of the tumor.

A few points need to be considered in future studies for better understanding of the presented results. Firstly, a better understanding is needed of iron oxide distribution in the center of the tumor xenograft. Secondly, the iron content in tumors as a function of time should be quantified in future studies by inductively coupled plasma atomic emission spectrometry. In addition, traces of iron oxide were found in the tumor periphery but they were not identified by MR imaging. Future studies should aim to develop methods able to detect low concentrations of iron oxide by MRI [7, 24-27], such as the development of highly sensitive MR contrast agents or pulse sequences.

In summary, we demonstrated that the label and inoculation method is feasible, MR imaging can monitor the kinetics of SPIO distribution following the development of human pancreatic cancer xenografts in nude mice, and iron was mainly distributed in the center of the tumor xenograft, which caused hypointensity on T1- and T2-weighted and 3D FIESTA images.

4.1 Conflict of Interest

All authors declare no possible conflicts of interest.

Acknowledgment

This work was supported by grant no. 30770612 from the National Nature Science Foundation of China.