MR Imaging Properties of ex vivo Common Marmoset Brain after Formaldehyde Fixation.

PURPOSE: Ex vivo brains have different MRI properties than in vivo brains because of chemical changes caused by fixative solutions, which change the signal intensity and/or tissue contrast on MR images. In this study, we investigated and compared the MRI properties of in vivo and ex vivo brains.
METHODS: Using a Bruker 9.4T experimental scanner unit for animals (Biospin GmbH, Ettlingen, Germany), we performed this study on the common marmoset. We measured the relaxation and diffusion values in the white matter and cortex of common marmosets and compared these values between in vivo brains (n = 20) and ex vivo brains (n = 20). Additionally, we observed the relationship between the tissue fixation duration and MRI properties by imaging a brain that underwent long-term fixation in a preliminary examination (n = 1).
RESULTS: The T1 values of ex vivo brains were decreased compared with those of in vivo brains; however, there were no significant difference in the T2 and T2* values of in vivo and ex vivo brains. Axial, radial, and mean diffusivity values of ex vivo brains decreased to approximately 65% and 52% of those of in vivo brains in the cortex and white matter, respectively. Conversely, fractional anisotropy values were not significantly different between in vivo and ex vivo brains.
CONCLUSION: The T1 values and diffusion coefficient values of the ex vivo brains were strikingly different than those of the in vivo brains. Conversely, there were no significant changes in the T2, T2* or fractional anisotropy values. Altogether, the dehydration caused by tissue fixation and the reduction in brain temperature were involved in changing the relaxation and diffusion coefficient values. Here, it was difficult to specify all factors causing these changes. Further detailed study is needed to examine changes in MRI properties.

Introduction

Imaging brain specimens with MRI (i.e., ex vivo MRI) allows us to obtain higher-resolution images than those constrained by in vivo imaging because it enables image acquisition over periods as long as several days.[1] It is also useful in the study of brain anatomy because specimens can be subjected to section preparation for histological examination.[2]

Ex vivo brain MRI data are widely used in pathological and neurological studies.[3-6] For example, ex vivo brains have been used to evaluate changes in the white matter and/or volume changes in the hippocampus in Alzheimer’s disease, to measure the volume of the frontal lobe gray matter and/or the lateral ventricles in schizophrenia, and to determine changes that occur in multiple sclerosis.[4,7-11] The ex vivo brain is also useful in forensic neurology research.[12]

Several previous MRI studies have measured physical values,[13] such as relaxation and diffusion values, that reflect tissue conditions for in vivo tissue assessment. Experiments to measure these values and compare them between in vivo and ex vivo brains have been conducted using the brains of mice, macaques, and humans.[14-23] Thus, the MRI properties of ex vivo brains have been examined using specimens from various animal species. However, there have been few studies with a statistically sufficient number of animals. In addition, among the various animals, there are not many studies that have compared measurements of these values between in vivo and ex vivo brains of the common marmoset (Callithrix jacchus). The common marmoset belongs to the primate group. The body length of an adult common marmoset measures up to 25 cm from the neck to the tail and they weigh approximately 350–450 g in captivity.[24-26] The common marmoset is useful as a psychiatric/neurological disease model because the pathology is similar to that in humans. Because of its short gestation period, it is also useful for tracking hereditary tendencies. Therefore, they have been used in recent neuroscience research.[27,28] In addition, neuroscience studies involving the common marmoset have used transgenic (Tg) common marmosets as subjects. Tg mice have been conventionally selected for several studies, but this raises the issues of genetic and functional differences between mice and humans. The Tg common marmoset is considered to solve this problem.[29,30] In fact, this animal was selected as a model in a Japanese national research project, Brain Mapping by Integrated Neurotechnologies for Disease Studies (Brain/MINDS).[31,32] This project aims to establish a basis for elucidation of the structure and function of the human brain to help develop new treatments for psychiatric and neurological disorders using common marmosets as an animal model.[33,34]

In this study, we aimed to understand the MRI properties of in vivo and ex vivo brains of the common marmoset. We assessed differences in relaxation and diffusion values between in vivo and ex vivo brain MRI data. Furthermore, as a preliminary examination, we examined the relationships between fixation duration and MRI properties by imaging a brain that underwent long-term fixation (n = 1) as tissues of some specimens are often fixed and preserved for long periods of time.

Materials and Methods

Animals

This study was approved by the Animal Experiment Committees at the RIKEN Brain Science Institute and was conducted in accordance with the Guidelines for Conducting Animal Experiments of the RIKEN Brain Science Institute (H27-2-307).

Twenty healthy common marmosets (mean age, 6.0 ± 2.1 years; sex, 8 males and 12 females) and 20 ex vivo brains (fixed for 2.4 ± 0.9 days) were included in this experiment.

In a preliminary examination, the brain of a healthy 4-year-old male common marmoset was scanned (1 time in vivo and 11 times ex vivo).

Magnetic resonance imaging

MRI was performed using a 9.4T BioSpec 94/30 (Biospin GmbH, Ettlingen, Germany) unit and a transmitting and receiving coil with an 86-mm inner diameter (40 mm for ex vivo brains). We obtained T1, T2, and mappings and diffusion-weighted images (DWI) from each animal. For T1 mapping, rapid acquisition with relaxation enhancement was used with the following parameters: TR = 1200/1600/3200/4800/10000 ms, TE = 7 ms, flip angle = 90°, number of averages (NA) = 1, and scan time = 20 min. For T2 mapping, a multiple spin-echo sequence was used with the following parameters: TR = 7000 ms, TE = 8/16/24/32/40/48 ms, flip angle = 90°, NA = 2, and scan time = 15 min. For mapping, a multiple gradient-echo sequence was used with the following parameters: TR = 2000 ms, TE = 3.5/8.5/13.5/18.5/23.5/28.5/33.5/38.5/43.5 ms, flip angle = 60°, NA = 2, and scan time = 10 min. The resolution was set at 270 × 270 × 540 μm in all cases. A partial coronal section perpendicular to the anterior comisure-posterior comisure (AC–PC) line and centered on PC was scanned in consideration of the limit of imaging setting. For DWI, spin-echo imaging and echo-planar imaging were used for the assessment of diffusion properties with the following parameters: TR = 3000 ms, TE = 25.57 ms, resolution = 350 × 350 × 700 μm, δ = 6 ms, Δ = 12 ms, b-value = 1000 s/mm2 in 30 diffusion directions (plus 2 b0 images), NA = 3, and scan time = 30 min. Our in vivo and ex vivo brain experiments were performed under the same measurement conditions to permit accurate statistical analysis of the differences in relaxation and diffusion values between in vivo and ex vivo brains.

To acquire in vivo brain data, the animals were scanned in the supine position on an imaging stretcher and administered a mixture of oxygen and 1.5–2.5% concentrated isoflurane (Abbott Laboratories, Abbott Park, IL, USA). During the scan, heart rate, peripheral oxygen saturation (SpO2), respiration, and rectal temperature were monitored regularly to manage the animal’s physical condition. Ex vivo brains were also obtained by the same perfusion procedure but scanned after 2–3 days of fixation. To acquire ex vivo brain data, the brain was wrapped in a sponge and soaked in a fluorine solution, which exhibits no signal on MRI, in a plastic container. Vacuum degassing was performed to reduce air bubble-derived artifacts.

Data analysis

The MRI properties in the current study were defined as physical values measured by mapping the relaxation time, i.e., T1, T2, and mapping, as well as tensor-based diffusion properties, i.e., fractional anisotropy (FA), mean diffusivity (MD), axial diffusivity (AD), and radial diffusivity (RD). These properties were calculated with ParaVision 6.0.1 (Bruker, Inc., Ettlingen, Germany).

For image preprocessing, in vivo brain data were subjected to digital skull stripping (isolating the cerebral parenchyma) using Amira version 6.0 (Visage Imaging, Inc., San Diego, CA, USA). Since it was not possible to scan the whole brain with the maximum slice number at the shortest TR in the multi-slice method, the T1-weighted images (T1WI), T2-weighted images (T2WI), and -weighted images ( WI) were obtained by scanning a partial coronal section perpendicular to the AC–PC line and centered on PC. It is unified with all subjects and imaging dates. The diffusion images were registered to a “standard brain” image[27] using an Advanced Normalization Tools (ANTs) open-source software script.[35] Thus, the images were in the same space and the same ROIs could be used to generate the values of the MRI properties regardless of image acquisition point in space, day, or animal.

To assess the values of the MRI properties, ROIs were selected and drawn. The ROIs were 2-dimensional and drawn in the parasagittal cortex and white matter near the vertex (Fig. 1). We used hand-placed ROIs to prevent errors from including the margins of each region. We measured each value in both brain hemispheres and calculated the average. The images were measured in the coronal plane with ImageJ software version 1.5.1 (National Institutes of Health, Bethesda, MD, USA). The normality of the data for each value was confirmed using the Jarque–Bera test. The in vivo and ex vivo data were then statistically compared using a Student’s t-test with Excel for Mac version 16.17 (Microsoft, Redmond, WA, USA). The level of significance was defined as P < 0.05.

Fig. 1

A diagram showing the ROIs drawn in the brain. This image shows the ROIs (circle: cortex, dotted circle: white matter) drawn for measurements.

Results

The data from each set of 20 marmosets obtained by ROI analysis were used to draw box plots and examine differences between in vivo and ex vivo brains (Fig. 2). The T1 relaxation values were significantly different between in vivo and ex vivo brains (Fig. 2a). A significant difference was also detected for the T2 relaxation values in the white matter but not in the cortex (Fig. 2b). Additionally, a significant difference was detected for the relaxation values in the cortex but not in the white matter (Fig. 2c). AD, RD, and MD values were significantly different between in vivo and ex vivo brains (Fig. 2e–2g). In the cortex, the mean of the diffusion coefficient values from ex vivo brains were 65% of that from in vivo brains. In the white matter, the mean of the values from ex vivo brains were 52% of that from in vivo brains. There were no significant differences in FA values between in vivo and ex vivo brains in the two regions (Fig. 2d). Figure 3 shows a color map of the relaxation and diffusion values of in vivo and ex vivo brains. Figure 4 shows the relationship between the values and the postmortem duration in the preliminary examination plotted using a line graph. The T1 values in these regions decreased remarkably within 3 days after tissue fixation and then showed a gradual decrease (Fig. 4a). The T2 and values in these regions decreased remarkably within 1 week after tissue fixation and then showed a gradual decrease similar to the trend seen in T1 values (Fig. 4b–4c). The AD, RD, and MD values in these regions showed a large decrease immediately after tissue fixation and remained approximately constant thereafter (Fig. 4e–4g). Regarding FA values, a significant change was not observed in these regions during the observation period (Fig. 4d).

Fig. 2

MRI measurement values from in vivo brains of 20 common marmosets and ex vivo brains from another set of 20 common marmosets. The top panel shows the tracking results of the changes in the relaxation values, i.e., T1, T2, and from the left. The bottom panel shows the results of FA, AD, RD, and MD from the left. The left and right halves of each figure show the values in the cortex and white matter (WM), respectively. The gray and white boxes represent the cortex and WM, respectively. The P-value listed at the top of each box plot was calculated by t-test. The mean and standard deviation are listed at the bottom of each box plot. AD, axial diffusivity; FA, fractional anisotropy; MD, mean diffusivity; RA, radial diffusivity.

Fig. 3

The difference of the relaxation values and the diffusion values of in vivo and ex vivo brain. The top panels show the results of the T1, T2, , FA, AD, RD, and MD maps of an in vivo brain from the left. The bottom panels show the results of those maps of an ex vivo brain from the left. The color bar for each map is displayed at the bottom of each map image. AD, axial diffusivity; FA, fractional anisotropy; MD, mean diffusivity; RA, radial diffusivity.

Fig. 4

Long-term measurements of magnetic resonance imaging values in the ex vivo marmoset brain. The top panel shows changes in values related to relaxation properties, i.e., T1, T2, and . The bottom panel shows FA, AD, RD, and MD results from the left. The curves in the graphs plotted with “circle” and “x” data points show the changes in the cortex and white matter, respectively. AD, axial diffusivity; FA, fractional anisotropy; MD, mean diffusivity; RA, radial diffusivity.

Discussion

Our data indicated that T1 and the diffusion coefficient values greatly changed between in vivo and ex vivo brains while FA values were unchanged without regard for the brain state. The changes in relaxation values may be caused by the effects of tissue fixation and the reduction in brain temperature. The effects of tissue fixation were reported by Thavarajah et al.[36] in detail. The paraformaldehyde solution used in this study induces a cross-linking reaction between the functional groups of macromolecules such as proteins, which are the main components of brain tissue. The cross-linking caused by the tissue fixative solution occurs because of a chemical reaction. This maintains the protein and carbohydrate structures and prevents tissue autolysis and decay. At the same time, dehydration of the specimen tissue occurs. In addition, Birkl et al.[37] noted that the major reasons for the changes in the relaxation values were not only dehydration but also the decrease in brain temperature. They also showed that T1 values were more affected by the change in brain temperature than were the T2 and values. This tendency is consistent with the results from this study. In this study, the in vivo brain temperatures of the marmosets subjected to the experiment were speculated to be approximately 38°C, in reference to the study by Hayward and Baker.[38] In contrast, the ex vivo brain was maintained at room temperature at approximately 20°C during MRI scanning. Therefore, there was an approximately 20°C difference between the in vivo and ex vivo brains.

There is a high possibility that this temperature difference had a large effect on the relaxation value results. In summary, dehydration caused by tissue fixation and brain temperature reduction may have caused the reduction in relaxation values observed in our study. However, it is also clear that other factors affected the relaxation values and further examination is necessary in the future.

Research by D’Arceuil et al.[17] and Holz et al.[39] showed that decreasing brain temperature decreased diffusion coefficient values. Using the relational equation between the absolute temperature and diffusion coefficient derived by Holz et al.,[39] the diffusion coefficient value at 20°C is about 65.2% of the value at 38°C. The AD, RD, and MD values of the ex vivo brain were 66.7%, 64.3%, and 66.2% of those from the in vivo brain in the cortex and 53.1%, 50.0%, and 52.1% in the white matter, respectively. Accordingly, the differences in these values in the cortex were almost equal to the values calculated by the estimation equation shown in the previous study. On the contrary, in the white matter, the values in this study were about 10% lower than the values calculated by the estimation equation. Factors, other than dehydration, arising due to tissue fixation and temperature change may be involved in the difference between the cortex and white matter. However, we could not clarify the specific factors in this research and a more detailed examination is necessary.

On the contrary, our research showed that the FA values did not change after tissue fixation. This result is similar to the results shown in previous studies by D’Arceuil et al.[17] and Guilfoyle et al.[18] The FA values are calculated from diffusion coefficient values in each direction (λ = 1–3) and are unchanged when the values in all directions decrease by the same ratio. As described above, the rates of reduction in AD (λ = 1) and RD (average values of λ = 2 and 3) were almost the same in both regions. Therefore, there were no significant differences between the FA values of in vivo and ex vivo brains.

As a result of the long-term tissue fixation conducted in a preliminary study, we infer that structural MR images, such as T1WI and T2WI, will be nearly the same 1 week after tissue fixation. Similarly, it is obvious that the FA map can be obtained regardless of the tissue fixation duration. However, it should be noted that the diffusion coefficient values change significantly before and after tissue fixation treatment. Generally, tissue fixation is performed for about 2–3 days when performing a pathological assessment. However, in preclinical studies, we occasionally deal with tissue specimens that have been fixed for a long time. Therefore, experimental data from long-term fixed tissue serve as a reference for examination of MRI conditions and the interpretation of measured values when such specimens are imaged.

The limitations of this study need to be acknowledged. First, the brain temperature was not measured directly because of technical problems. In addition, in order to evaluate the factors causing changes in the relaxation and diffusion coefficient values in more detail, it is necessary to evaluate the existence of structural denaturation. However, in vivo micro imaging at the pathological level is difficult technically, so this examination could not be performed in this study.

A strength of our study is that we examined MRI properties using a statistically sufficient number of animals. However, additional research is required to clarify other factors that caused changes in each value. The next step would be to analyze similarities in nerve structures between in vivo and ex vivo brains. In this study, we showed that there were no significant differences between the FA values of in vivo and ex vivo brains. On the contrary, according to a previous study, the length and density of nerve fibers decreased with time after death.[40] Therefore, it will be important to look for differences in tractography results from in vivo and ex vivo brains to separate the influence of fixation time from that of diffusion time. If it is possible to evaluate neural structures equally in in vivo and ex vivo brains, a study on ex vivo brains with no limitation on the MRI acquisition duration would be highly useful.

Conclusion

In this study, we measured relaxation and diffusion values and compared them between in vivo and ex vivo brains. The T1 values of the ex vivo brains decreased to 80% of those of the in vivo brains. On the contrary, no significant changes in the T2 or values were observed between the in vivo and ex vivo brains. In addition, the diffusion coefficient values of the in vivo and ex vivo brains were significantly different in both the white matter and cortex regions. This decrease occurred at a roughly constant rate and the FA values were not significantly different between in vivo and ex vivo brains. We infer that the dehydration caused by tissue fixation and the reduction in brain temperature were related to the change in each value. However, since it is difficult to identify all factors affecting these changes, due to technical limitations, further studies are needed to investigate changes in MRI properties in greater detail.