Novel In Vivo Micro-Computed Tomography Imaging Techniques for Assessing the Progression of Non-Alcoholic Fatty Liver Disease

Summary

Using a diet-induced non-alcoholic fatty liver disease (NAFLD) mouse model, we describe the use of novel in vivo micro-computed tomography imaging techniques as a non-invasive method to assess the progression stages of NAFLD, focusing predominantly on the hepatic vascular network due to its significant involvement in NAFLD-related hepatic dysregulation.

Abstract

Non-alcoholic fatty liver disease (NAFLD) is a growing global health issue, and the impact of NAFLD is compounded by the current lack of effective treatments. Considerable limiting factors hindering the timely and accurate diagnosis (including grading) and monitoring of NAFLD, as well as the development of potential therapies, are the current inadequacies in the characterization of the hepatic microenvironment structure and the scoring of the disease stage in a spatiotemporal and non-invasive manner. Using a diet-induced NAFLD mouse model, we investigated the use of in vivo micro-computed tomography (CT) imaging techniques as a non-invasive method to assess the progression stages of NAFLD, focusing predominantly on the hepatic vascular network due to its significant involvement in NAFLD-related hepatic dysregulation. This imaging methodology allows for longitudinal analysis of liver steatosis and functional tissue uptake, as well as the evaluation of the relative blood volume, portal vein diameter, and density of the vascular network. Understanding the adaptations of the hepatic vascular network during NAFLD progression and correlating this with other ways of characterizing the disease progression (steatosis, inflammation, fibrosis) using the proposed method can pave the way toward the establishment of new, more efficient, and reproducible approaches for NAFLD research in mice. This protocol is also expected to upgrade the value of preclinical animal models for investigating the development of novel therapies against disease progression.

Introduction

Non-alcoholic fatty liver disease (NAFLD) is a metabolic disease that affects approximately 25% of the population and >80% of morbidly obese people¹. An estimated one-third of these individuals progress to non-alcoholic steatohepatitis (NASH), which is characterized by hepatic steatosis, inflammation, and fibrosis². NASH is a disease stage with a significantly higher risk for the development of cirrhosis and hepatocellular carcinoma (HCC)^3,4. For this reason, NASH is currently the second most common cause of liver transplantation, and it is also expected to soon become the most important predictor of liver transplantation^5,6,7. Despite its prevalence and severity, no disease-specific therapy is available for NAFLD, and the existing treatments only aim to tackle disease-associated pathologies such as insulin resistance and hyperlipidemia^5,6.

In recent years, the pathophysiological role and adaptations of the endothelium and, in general, of the vascular network of metabolic tissues, such as the adipose tissue and the liver, have been gaining more importance in research, especially during obesity and metabolic dysregulation^7,8. The endothelium is a cellular monolayer that lines the vascular network internally, acting as a functional and structural barrier. It also contributes to various physiological and pathological processes, such as thrombosis, metabolite transport, inflammation, and angiogenesis^9,10. In the case of the liver, the vascular network is, among other features, characterized by the presence of highly specialized cells, defined as liver sinusoidal endothelial cells (LSECs). These cells lack a basement membrane and have multiple fenestrae, allowing for the easier transfer of substrates between the blood and liver parenchyma. Due to their distinctive anatomical location and characteristics, LSECs likely have a crucial role in the pathophysiological processes of the liver, including the development of liver inflammation and fibrosis during NAFLD/NASH. Indeed, the pathological, molecular, and cellular adaptations that LSECs undergo in the course of NAFLD contribute to the disease progression¹¹. Specifically, the LSEC-dependent hepatic angiogenesis that takes place during NAFLD is significantly associated with the development of inflammation and the progression of the disease to NASH or even HCC¹². Besides, obesity-related early NAFLD is characterized by the development of insulin resistance in LSECs, which precedes the development of hepatic inflammation or other advanced NAFLD signs¹³.

Additionally, LSECs have recently emerged as central regulators of hepatic blood flow and vascular network adaptations during liver disease of several etiologies^14,15. Indeed, chronic liver disease is characterized by prominent intra-hepatic vasoconstriction and increased resistance to blood flow, which contribute to the development of portal hypertension¹⁶. In the case of NAFLD, several LSEC-related mechanisms contribute to this phenomenon. For instance, LSEC-specific insulin resistance, as mentioned above, is associated with reduced insulin-dependent vasodilation of the hepatic vasculature¹³. Besides, over the course of the disease, the liver vasculature becomes more sensitive to vasoconstrictors, further contributing to impairment of the hepatic blood flow and leading to the emergence of shear stress, which both result in a disruption of the sinusoidal microcirculation¹⁷. These facts suggest that the vasculature is a key target in liver disease. Nevertheless, limiting factors hindering the timely diagnosis and monitoring of NAFLD/NASH, as well as the development of potential therapies, are the inadequacies in the consistent characterization of the hepatic microenvironment and (micro)vascular structure, as well as the scoring of the disease stage in a spatiotemporal and non-invasive manner.

Micro-computed tomography (CT) imaging is currently the gold-standard non-invasive imaging method for accurately depicting anatomical information within a living organism. Micro-CT and MRI represent two complementary imaging methods that can cover a vast range of pathologies and provide exceptional resolution and detail in the imaged structures and tissues. Micro-CT, in particular, is a very fast and accurate tool that is often used for studying pathologies such as bone diseases and associated bone surface changes¹⁸, assessing the progression of pulmonary fibrosis over time¹⁹, diagnosing lung cancer and its staging²⁰, or even examining dental pathologies²¹, without any special preparation (or destruction) of the samples being imaged.

The imaging technology of micro-CT is based on the different attenuation properties of various organs in terms of the interaction of X-rays with matter. Organs presenting high X-ray attenuation differences are depicted with high contrast in CT images (i.e., the lungs appear dark and the bones light). Organs presenting very similar attenuation properties (different soft tissues), are challenging to distinguish on CT images²². To address this limitation, specialized contrast agents based on iodine, gold, and bismuth have been extensively investigated for in vivo use. These agents alter the attenuation properties of the tissues in which they accumulate, are cleared slowly from the circulation, and enable the uniform and stable opacification of the entire vascular system or chosen tissues²³.

In human diagnostics, CT imaging and comparable techniques, such as MRI-derived proton density fat fraction, are already in use for the determination of hepatic fat content^24,25. In the context of NAFLD, high soft tissue contrast is essential to accurately distinguish pathological lesions or small vessels. For this purpose, contrast agents providing enhanced contrast of the liver tissue characteristics are utilized. Such tools and materials allow for the study of multiple liver characteristics and possible pathology expressions, such as the architecture and density of the vascular network, lipid deposition/steatosis, and functional tissue uptake/lipid (chylomicron) transfer in the liver. Additionally, hepatic relative blood volume and portal vein diameter can also be evaluated. In a very short scan time, all these parameters provide different and complementary information on the evaluation and progression of NAFLD, which can be used to develop a non-invasive and detailed diagnosis.

In this article, we provide a step-by-step protocol for the use of novel in vivo micro-CT imaging techniques as a non-invasive method to assess the progression stages of NAFLD. Using this protocol, the longitudinal analysis of liver steatosis and functional tissue uptake, as well as the evaluation of the relative blood volume, portal vein diameter, and density of the vascular network, can be performed and applied in mouse models of liver disease.

Protocol

All procedures were carried out by BIOEMTECH's personnel in accordance with European and national welfare regulations and were approved by national authorities (license number EL 25 BIOexp 45/PN 49553 21/01/20). All experiments were designed and reported with adherence to ARRIVE guidelines²⁶. The mice were purchased from the Hellenic Pasteur Institute, Athens, Greece.

NOTE: Animals were group-housed in individually ventilated cages enriched with rails and cardboard tubes in a room at 20-22 °C, with a relative humidity of 50%-60% and a 12 h light/dark cycle (light 07:00 am-07:00 pm). A combination of a high-fat diet (HFD) and high-fructose corn syrup (HFCS), a fructose- and glucose-containing sweetener widely used in modern types of fat-enriched diets, was used to induce NAFLD as a recognized reliable model^27,28,29,30. At 7-8 weeks of age, male C57BL/6 mice were given ad libitum access to either a normal diet (n = 2) with 10% of kilocalories from fat or a HFD (n = 2) containing 60% of kilocalories from fat supplemented with 5% HFCS in water for 22 weeks. Body weight was obtained weekly using a digital balance, and during the experimental period, animal welfare was monitored on alternate days using a score sheet. At the end of the imaging protocol, the mice were euthanized via cervical dislocation.

1. Animal preparation

NOTE: The imaging protocol is summarized in Figure 1.

1. Anesthetize the mouse using 3%-4% isoflurane (in room air), and maintain its body temperature using a dedicated heating pad.
   NOTE: The absence of a pedal withdrawal reflex must be used to confirm sufficient anesthesia depth before initiating the scan.
2. Apply ophthalmic ointment on the animal's eyes prior to experimentation.
3. Place the animal in the CT scanner cradle, secure the nose cone, and switch to 1.5%-3% isoflurane (in room air) for maintenance.
   NOTE: The absence of a pedal withdrawal reflex must be used to confirm the appropriate percentage of isoflurane for the maintenance of anesthesia.
4. Monitor the mouse continuously.

2. Pre-scanning preparation

NOTE: Imaging is performed in two experimental phases to allow for the first contrast agent to be adequately cleared from the circulation and tissues. eXIA (first contrast agent) is administered in the first phase and ExiTron (second contrast agent) in the second phase, as described in the "Imaging workflow" section (section 3) below.

1. Allow for the contrast agent (either eXIA or ExiTron, depending on the experimental phase) to reach room temperature for 3 h.
2. Set the following scanning parameters on the CT scanner: high-resolution protocol under 50 kVp tube voltage and a current of 460 µA, non-spiral, 720 projections/rotation, four rotations, and 4 min acquisition time.

3. Imaging workflow

1. Experimental phase 1
   1. Calculate and prepare the volume of the first contrast agent to be administered at an undiluted dose of 6 µL/g of body weight for maximum contrast.
   2. Prepare the tail vein catheter by filling it up with saline and connecting it to the syringe filled with the contrast agent.
   3. Acquire a pre-contrast whole body (WB) and liver baseline scan.
   4. Ensure that there are no bubbles or blockages in the syringe or catheter.
   5. Insert the prefilled catheter into the tail vein, and administer the contrast agent via an injection performed slowly and manually, with a duration of 1-3 min (not as a bolus injection). A syringe pump can be used when set to the appropriate infusion rate.
      NOTE: The tail of the animal can be placed in lukewarm water to induce vasodilation and help with the catheter insertion
   6. Acquire WB and liver scans at different time points, as indicated in Table 1.
      NOTE: If the acquisition of all points is not possible, the focus should be placed on 45 min post-injection (PI), which is the point of maximum liver uptake, and 48 h PI, which is when clearance is achieved.
2. Experimental phase 2
   1. Prepare the mouse again as described in section 1 for the administration of the second contrast agent 10 days following the final reading with the first contrast agent (48 h PI).
   2. Perform steps 2.1-2.2.
   3. Calculate and prepare the volume of the second contrast agent to be administered at an undiluted dose of 8 µL/g of body weight for maximum contrast.
   4. Prepare the tail vein catheter by filling it up with saline and connecting it to the syringe filled with the contrast agent.
   5. Acquire a pre-contrast WB and liver baseline scan to evaluate the relative blood volume and liver steatosis.
   6. Ensure no contrast is detectable in the scan as an indication of the complete clearance of the first contrast agent.
   7. Insert the prefilled catheter into the tail vein, and administer the contrast agent via an intravenous injection performed slowly and manually, with a duration of 1-3 min (not as a bolus injection). A syringe pump can be used when set to the appropriate infusion rate.
   8. Acquire WB and liver scans at different time points, as indicated in Table 1.
      ​NOTE: WB scans are acquired at 10 min and 4 h PI. The significant time lapse between them allows for the evaluation of the tracer biodistribution in the body as well as its relative clearance.

4. Data extraction and analysis

NOTE: In this protocol, the data extraction and analysis steps based on a specific imaging processing software (see Table of Materials) are provided. The described steps may need to be adapted when using different software.

1. Evaluation of hepatic lipid deposition/steatosis.
   NOTE: For the evaluation of hepatic steatosis, no contrast agent is used, and a comparison between control and pathology is performed. Due to relatively high deviations in tissue attenuation properties between different mice, the density values are normalized for the liver against the spleen (fat-free tissue) and the fat (absolute fat tissues) according to the following equation and as previously described²⁵:
   
   1. To perform the analysis, load the DICOM file of the pre-contrast scan, and adjust the bar/contrast to see the liver, spleen, and white adipose tissue (WAT) clearly.
   2. Access the Modelling Operator tool via the tool pull-down menu on the front panel, and select 3D ROI Tool.
   3. Under the 3D ROI Operator, select Add ROI to generate multiple ROIs (up to eight for each tissue) to perform sampling in the areas where the liver (preferably in the areas of the left medial lobe, the right medial lobe, and the left lateral lobe) and spleen appear clear, with no apparent blood vessels and fat.
      NOTE: For WAT, ROIs are selected in the middle of the visceral adipose tissue depot. Recommended areas are shown in Figure 2. Normalization methods using the liver/spleen ratio and without including WAT can also be applied as previously established³¹.
   4. Under the 3D Paint Mode and Erode/Dilate feature, select 2D, and use the interface that appears to specify a name and color for each ROI.
   5. Use the Sphere paint ROI tool with a diameter of 8 pixels to manually draw the 2D ROIs.
   6. Perform sampling by segmenting the 2D ROIs on the areas of interest using the Cross Hair tool on the transverse plane, as shown in Figure 3A.
   7. Click on the selected point at the sagittal and coronal plane to complete the segmentation of the 2D ROI, as shown in Figure 3B.
   8. Repeat the process for defining the rest of the ROIs.
      NOTE: When sampling, avoid the organ border regions, as this can introduce noise and affect the reliability of the calculated Hounsfield unit (HU) value of each ROI.
   9. Once satisfied with the segmented ROIs, go to Navigation, and select Show Table to display the quantification table containing the calculated HU values for each ROI.
      NOTE: The values of interest are listed in the "Mean" column, which contains the numerical mean values of the voxels (HU) contained in the ROIs for the organs of interest. Note the values of interest, or save the entire table by selecting Export Table.
   10. Calculate the average HU for the liver, spleen, and WAT, and plug the values into the above equation to calculate the percentage of liver fat.
2. Functional tissue uptake/lipid (chylomicron) transfer in the liver
   NOTE: Functional tissue uptake/lipid (chylomicron) transfer is analyzed from the acquired scans at 45 min and 48 h following the first contrast agent infusion, based on a previously published method³². The contrast is calculated for the different tissues and time points using the equation below:
   
   PV organ t_i is the average pixel value in the organ at time t_i (ranging from 0 h to 48 h), and PV organ t₀ is the average pixel value in the organ in the image without contrast.
   1. To perform this analysis, load the eXIA scan DICOM file, and adjust the bar/contrast to see the liver, spleen, and left ventricle clearly.
   2. Access the Modelling Operator via the tool pull-down menu on the front panel, and select 3D ROI Tool.
   3. Under the 3D ROI Operator, select Add ROI to segment multiple ROIs for the liver.
   4. Under the 3D Paint Mode and Erode/Dilate feature, use the Sphere paint ROI tool with a diameter of 8 pixels and −1 erode.
      NOTE: Select multiple ROIs on slices where each organ appears clearly. Avoid border regions, as this can introduce noise and affect the reliability of the calculated HU value of each ROI. This will result in the sampling of multiple 3D ROIs, which correspond to small organ volumes.
   5. Use the interface that appears to specify a name and color for each ROI.
   6. Once satisfied with the designed ROIs, go to Navigation, and select Show Table to display the quantification table containing the calculated HU values for each ROI.
      NOTE: The values of interest are listed under the "Mean" column, which displays the numerical mean values of the voxels (HU) included in the ROI. The average HU value of each organ's ROIs corresponds to PV organ t_i. Note the values of interest, or save the entire table by selecting Export Table.
   7. To obtain PV organ t₀, repeat all of the above steps using the pre-contrast DICOM file to calculate the mean brightness of the liver, spleen, and left ventricle before the contrast agent injection.
   8. Insert the values into the above equation to extract the percentage contrast corresponding to the functional tissue uptake/lipid (chylomicron) transfer.
3. Architecture and density of the hepatic vascular network
   NOTE: The analysis of the architecture and density of the hepatic vascular network is based on a previously published methodology³³ and is performed on the liver scans obtained 10 min PI of the second contrast agent.
   1. To perform this analysis, load the ExiTron scan DICOM file, and adjust the bar/contrast to see the liver vascular network clearly.
   2. Access the Modelling Operator via the tool pull-down menu on the front panel, and select 3D ROI Tool.
   3. Under the 3D ROI Operator, select Add ROI to generate a 3D ROI for the liver.
   4. Under the 3D Paint Mode and Erode/Dilate feature, select 3D.
      NOTE: Use the Sphere paint ROI tool with −1 erode to define the segmentation layers across the coronal plane. The diameter of the ROI painting tool must be adjusted according to each layer (for adding/deleting any wanted/unwanted voxel selections). It is recommended that the liver volume be initially defined across the coronal plane, and then the transverse and sagittal planes can be used to correct the ROI. This process requires precision. The user must be very careful not to include other tissues, vessels, and bones when segmenting each ROI layer while ensuring that all areas of the liver are included in the defined ROI. For this reason, familiarization with the anatomical boundaries of the liver is crucial.
   5. Once satisfied with the resulting liver ROI, perform a cut in order to remove all the voxels from the image data that do not belong in the initially segmented liver ROI. For this, choose the liver ROI from the ROI Selector, and click on the Perform Cut icon. This operation removes the background and leaves the liver ROI unchanged.
      NOTE: Although the undo/redo functions are applicable to all the operations performed under the 3D ROI Tool, the action of cutting an ROI cannot be undone. So, prior to this action, the user might consider saving the initial liver ROI in DICOM format.
   6. The resulting liver ROI includes the vascular network and the surrounding tissue, which must be removed. For this, reset the liver ROI by clicking on the Reset ROI broom button.
   7. Use the interface that appears to transfer all the pixels of the liver ROI to the background.
      NOTE: The liver ROI will still exist after this operation, but it will no longer contain any voxels.
   8. To re-segment the liver ROI so that it only contains vascular-associated pixels, go to Segmentation Algorithms denoted by the magic wand icon, and select Connected Thresholding.
   9. Define the ROI as Output and the background as Input from the input drop-down menu before applying thresholding.
   10. Set the Thresholds by clicking the Min and Max icons to the left of each threshold field to fill in maximum and minimum values and obtain the vascular network.
       NOTE: Only pixels within the chosen range will be included in the resulting ROI. Adjusting the threshold values between different animals ensures that the same anatomical regions are taken into account with respect to the exact amount of contrast agent injected into each animal. This is constant among the selected tissues even if the numerical values are not identical.
   11. Use the Cross Hair tool to click on a point where the vascular network appears clear, and click on Apply to perform the segmentation.
   12. Activate the maximum intensity projection (MIP) viewer.
   13. Evaluate the resulting liver ROI in terms of how clear the vascular network appears in the MIP view.
   14. If the tissue remains in parts of the liver ROI, repeat steps 4.3.5-4.3.11 by adjusting the Min threshold value until the segmented liver ROI clearly represents the vascular network.
   15. Once satisfied with the resulting liver ROI, generate the quantification table containing the calculated liver ROI volume in cubic millimeters.
       NOTE: The values of interest are listed in the "mm³" column, which contains the numerical volume value of the voxels (HU) contained in the liver ROI. Note the values of interest, or save the entire table by selecting Export Table.
4. Hepatic relative blood volume
   NOTE: For the measurement of the hepatic relative blood volume (rBV), which correlates significantly with the amount of newly formed blood vessels during fibrosis progression, pre-contrast scans and scans at 4 h after the second contrast agent injection are used. The analysis is performed as previously described³⁴.
   1. To perform this analysis, load the ExiTron scan DICOM file, and adjust the bar/contrast.
      NOTE: Disable the MIP viewer: under Preferences, check the box to disable the MIP viewer upon loading. For large datasets, this can improve the loading speed.
   2. Access the Modelling Operator via the tool pull-down menu on the front panel, and select 3D ROI Tool. This tool provides advanced options for drawing, visualizing, saving, and quantifying both 2D and 3D regions.
   3. Under the 3D ROI Operator, select Add ROI, and segment two ROIs: one for the liver and one for a large blood vessel.
   4. Under the 3D Paint Mode and Erode/Dilate feature, select 2D.
      NOTE: It is recommended to use the Sphere paint ROI tool with a diameter of 8-10 pixels for the liver and 4-6 pixels for the blood vessel. However, the paint tool diameter can be adjusted depending on how small the area to be selected is.
   5. Use the interface that appears to specify a name and color for each ROI.
      NOTE: Select two to five slices of central parts of the tissues of interest, and define the segmentation layers to generate 2D ROIs for each tissue. When selecting the areas on each slice, avoid the organ border regions, as shown in Figure 4, as this can introduce noise and affect the reliability of the calculated HU value of each ROI.
   6. Once satisfied with the designed ROIs, go to Navigation, and select Show Table to display the quantification table containing the calculated HU values for each ROI.
      NOTE: The values of interest are listed under the "Mean" column, which displays the numerical mean values of the voxels (HU) included in the ROI. Note the values of interest, or save the entire table by selecting Export Table.
   7. Repeat all the steps for the pre-contrast DICOM file to obtain the mean brightness of the liver before the contrast agent injection. For this, carry out steps 4.4.2-4.4.5 for the liver only.
   8. Calculate the average HU values for each tissue at the equivalent time points, and insert the values obtained into the equation below:
      
      NOTE: A large blood vessel after contrast agent injection is considered 100% rBV, and the liver, before contrast agent administration, is considered 0% rBV.
5. Portal vein diameter
   NOTE: For the portal vein diameter measurements, the same scans used for the hepatic rBV measurements are analyzed as previously described³⁵.
   1. Load the ExiTron scan DICOM file, and adjust the bar/contrast.
   2. Locate the transversal planes of three to four slices above the junction of the superior mesenteric and splenic veins (Figure 5).
   3. Use the Ruler tool to measure the exact distance between two points (i.e., the diameter of the circular vein region).
      NOTE: The distance is extracted on the image, but one can also go to Navigation and select Show Table to display the quantification table containing the calculated distance or select Export Table to save the result.

Results

In this representative study, micro-CT imaging without any contrast agent indicated a higher percentage of liver fat in mice with NAFLD compared to controls (Table 2), confirming the pathology. Using the ExiTron contrast agent and the hepatic vascular network architecture and density analysis described above, the total volume density of the hepatic vascular network was found to be higher in mice with NAFLD compared to healthy controls (Figure 6, Table 2). Mice with NAFLD also had a larger portal vein diameter compared to control mice (Table 2), a structural alteration associated with portal hypertension during liver disease^34,36,37. Similarly, the hepatic rBV of animals with NAFLD was calculated to be higher compared to healthy controls (Table 2).

Furthermore, the analysis of the functional tissue uptake assay indicated a higher accumulation and slower clearance of the eXIA contrast agent in mice with NAFLD compared to healthy controls (Figure 7). These results suggest that steatotic hepatocytes likely undergo high levels of cellular endocytosis and distribution, though accompanied by reduced metabolic catabolism or clearance/secretion. Overall, this finding aligns with the phenotype of reduced hepatic metabolic activity, as expected in the case of fatty infiltration/NAFLD^38,39.

Figure 1: Schematic depicting an overview of the experimental protocol. Please click here to view a larger version of this figure.

Figure 2: Representative images highlighting the proposed areas used to perform sampling for segmenting the 2D ROI for the liver (red), spleen (green), and WAT (blue). Please click here to view a larger version of this figure.

Figure 3: Representative example of segmenting a 2D ROI for calculating steatosis. (A) The Cross Hair tool is used to select a 2D ROI on a desired area of the liver on the transverse plane. (B) The process is repeated on the sagittal and coronal planes to complete the segmentation of the 2D ROI. Please click here to view a larger version of this figure.

Figure 4: Representative example of a segmentation layer for generating a 2D liver ROI. Central parts of the organ are manually selected, avoiding the border regions to eliminate noise. Please click here to view a larger version of this figure.

Figure 5: 2D axial area. Representative examples of the 2D axial area of the different slices chosen to measure the portal vein diameter in mice with NAFLD and healthy controls. Please click here to view a larger version of this figure.

Figure 6: Hepatic vascular network architecture. Representative extracted images of the hepatic vascular network architecture, as shown through CT segmentation, obtained from a control mouse (176.9 mm³) and a mouse with NAFLD (390.3 mm³). Please click here to view a larger version of this figure.

Figure 7: Average contrast values. Grouped data showing the average contrast values, which represent the hepatic tissue uptake over 48 h post injection of eXIA in mice with NAFLD (n = 2) and healthy controls (n = 2). All grouped data are expressed as mean ± standard deviation. Please click here to view a larger version of this figure.

	ExiA		ExiTron
Time Point	Whole Body Scan	Liver Scan	Whole Body Scan	Liver Scan
Pre-contrast	X	X	X	X
10 min PI	-	-	-	X
15 min PI	O	O	-	-
45 min PI	X	X	-	-
2 h PI	O	O	-	-
4 h PI	-	-	X	X
24 h PI	O	O	-	-
48 h PI	X	X	-	-

Table 1: Time points of the CT scans. The appropriate time points of the whole body and liver scans acquired pre- and post-injection (PI) of the contrast agents. X indicates mandatory scans, - indicates no scans, and O indicates optional (recommended but not mandatory) scans.

	Control (n = 1–2)	NAFLD (n = 1–2)
% Liver fat	2.4 ± 1.5%	18.4 ± 3.1%
Hepatic vascular network volume	176.9 mm3	390.3 mm3
Portal vein diameter	1.1 mm	1.4 mm
Hepatic relative blood volume	~54%	~79%

Table 2: Representative results indicating differences in the percentage of liver fat, hepatic vascular network volume, portal vein diameter, and hepatic relative blood volume between mice with NAFLD and healthy controls.

Discussion

The current recommended method for NAFLD diagnosis and staging in humans is liver biopsy, which harbors the risk of bleeding complexities, as well as sampling inaccuracies⁴⁰. On the contrary, in animal models, such diagnosis is performed by histology post-mortem, although protocols for survivable liver biopsy are now available and are recommended when the study design allows⁴¹. The use of post-mortem histology means that a large number of animals are required to investigate the progression of this disease. As such, a study cannot be performed in the same animal while the disease progresses; variability is also expected to be higher when comparing samples obtained at different time points from different animals. In this article, we provide an innovative, non-invasive in vivo micro-CT approach, applied in an established experimental animal model of NAFLD, that enables the longitudinal evaluation of disease progression through the quantification of functional parameters, namely liver steatosis and functional tissue uptake, the relative blood volume, the portal vein diameter, and the density of the vascular network, in the same animal.

Micro-CT imaging is currently the gold-standard non-invasive imaging method for accurately depicting anatomical information within a living organism. Micro-CT has the capability to reach a high spatial resolution, thus representing a valuable tool for studying very fine details in a vast range of pathologies. Furthermore, it is a fast and reliable tool to study disease progression over time^21,22,23,24 by exploiting the intrinsic characteristics of X-rays without interfering with the integrity of the imaged sample. One major advantage of this technique in animal research is the ability to use a combination of selected contrast agents administered without any complicated preparations (non-invasive). This allows for several parameters to be extracted from the same animal scanned longitudinally during the progression of the disease, ensuring maximal output and compliance with the ARRIVE guidelines and the 3Rs²⁶. Furthermore, micro-CT scanners provide high-resolution images in very short acquisition times (with scan durations of just a few minutes or less), with the interpretation of the results in 2D and 3D formats being relatively easy (especially when using user-friendly software, as proposed here).

All the scans described in this protocol were performed on a small rodent CT scanner (see Table of Materials). The CT system performs a spiral scan, and it can provide images with 100 µm resolution. It operates between 35-80 kVp and 10-500 µA tube current. The CT data acquisitions last 7-10 min per scan and are reconstructed through an image space reconstruction algorithm (ISRA) at 100 µm spatial resolution. CT imaging is performed using the following parameters: i) high-resolution protocol at 50 kVp for WB scans and ii) high-resolution multirotation local scan at 50 kVp for local liver scans. All the parameters are set up in the scanning device software (see Table of Materials) provided with the system by following the instructions on the user interface. The reconstruction is performed through a Feldkamp, Davis, and Kress (FDK) algorithm with a voxel size of 0.1 mm. CT artifacts are minimized by periodically calibrating the detectors and maintaining the system well through appropriate services. If artifacts occur, the problematic detector that causes the ring artifact can be remotely canceled and the image corrected.

The described technical protocol was optimized using the eXIA and ExiTron contrast agents, which were selected due to their special formulations and temporal biokinetics in the various tissues. eXIA is a fully biodegradable contrast agent that contains 160 mg/mL iodine as an X-ray attenuating agent. Once injected intravenously, this contrast agent shows the blood residence time (with a clearance half-time of >30 min) and is then taken up by metabolically active organs such as the liver, spleen, myocardium, and brown adipose tissue. The first contrast agent is, therefore, appropriate for the non-invasive in vivo detection of liver and spleen abnormalities, myocardial infarction, and cardiomyopathy, as well as for identifying and quantifying active brown adipose tissue. ExiTron is an alkaline earth metal-based nanoparticle contrast agent of ~12,000 HU undiluted density⁴², which is specifically formulated for preclinical CT imaging. On intravenous injection, the second contrast agent circulates in the bloodstream, and it is taken up by the cells of the reticuloendothelial system⁴³, including macrophages within the liver.

There are certain limitations associated with the proposed methodology. For this protocol to be successful, no CT signal should be detected from the first contrast agent prior to administering the second contrast agent. The time frame proposed in this protocol was selected following optimization experiments (and based on other studies)⁴⁴, which suggested that the first contrast agent has slow clearance. As the second contrast agent has an even slower clearance rate^45,46, it must be administered second, following the complete clearance of the first contrast agent. Indeed, we have identified the clearance of the first contrast agent to be satisfactory after 12 days of injection. Depending on the duration of the animal model used, this time scale could allow the comparison of two time points at which the liver disease might have progressed. Considering that the feeding protocol used here takes 22 weeks, a time-lapse of 12 days is not expected to cause significant changes in the disease progression. The experimenter is required to perform appropriate validation and optimization before adjusting the proposed imaging protocol. One must also assess the cost-benefit analysis of the progression and image signal before altering the type and concentration of the contrast agents used, as well as the time frame of administration.

Additionally, both contrast agents can only be tolerated up to a specific total volume of administration before reaching toxic levels for the animal⁴³. Due to the fact that a maximal contrast agent volume is required to ensure optimal contrast, in combination with the slow clearance rate of the agents, a one-time injection is suggested for this analysis. Repeated administrations of these or other appropriate alternative contrast agents could also be used for different experiments, as long as the total volume of the contrast agent remains below the recommended limit at any given time point. Another requirement for successful imaging using this micro-CT protocol is the correct administration of the contrast agents. As stated in the protocol, the experimenter should ensure a slow infusion of the agent at the appropriate dosage directly into the bloodstream through the vein, without any bubbles. Failure to do so will compromise the imaging resolution and outputs. Therefore, the selection of an optimal dosage and appropriate administration methods (infusion and duration between different agents) reduces the risk of toxicity and the associated limitations in the CT imaging.

For scanning at multiple time points, attention should be given to keeping the anesthesia duration as short as possible. Since CT scanning only takes a few minutes to complete, the anesthesia can be reversed between some scans to minimize the risk associated with prolonged anesthesia durations. Repeated anesthesia inductions also have risks. However, placing the animals on a heated pad between scans and ensuring sufficient hydration helps with recovery and the maintenance of physiology. Furthermore, the exposure to ionizing radiation emitted by X-rays is adequate to influence organ and cell biology, making this potentially harmful to the animals and, consequently, resulting in biased and misleading experimental data²². As the expected dose to be delivered per scan is around 385 mGy, mice receiving multiple scans during the study can receive up to 1.8 Gy or more. This is a significant radiation dose for mice that could have potentially harmful effects on their tissue biology. This is especially concerning since an increase in dose is required when reducing the isotropic voxel spacing while maintaining the same image quality²².

In terms of image post-processing using the recommended software (see Table of Materials), segmentation masks are created using a mixture of region-growing and thresholding tools, and this is the most time-consuming step of the analysis. In some cases, manual modifications of the obtained segmentations should be carried out using a smoothing method. To optimize the edge-preserving and noise-reduction properties of the desired network, we recommend a Gaussian filter with a value of 0.3. Representative images of such a vascular network are shown in Figure 6 (post-processed using an open-access DICOM medical image viewer). The key limitation in terms of measuring the vascular network is that the software does not have the capability to accurately separate the selected defined ROI (which represents the vascular network of the liver) from the background (which represents the surrounding liver tissue); therefore, the appropriate threshold must be selected through trial and error. Initially, the user defines a lower threshold value of 600 HU and a maximum of 10,000 HU. If the extracted vascular network and the separation from the surrounding tissue are not acceptable, then the lower value is adjusted via trial and error following stepwise changes of 50-100 HU. The process is repeated by the user until the vascular network is sufficiently separated from the tissue.

In conclusion, understanding the adaptations of the hepatic vascular network during NAFLD progression and correlating them with other methods of disease characterization using the proposed method can pave the way towards the establishment of new, more efficient, and reproducible approaches for NAFLD research in mice. This protocol is also expected to upgrade the value of preclinical animal models for investigating the development of novel therapies against disease progression.

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
eXIA160	Binitio Biomedical, Inc.		https://www.binitio.com/?Page=Products
High fat diet with 60% of kilocalories from fat	Research Diets, New Brunswick, NJ, USA	D12492
High-fructose corn syrup	Best flavors, CA	hfcs-1gallon
Lacrinorm ophthalmic ointment	Bausch & Lomb
Normal diet with 10% of kilocalories from fat	Research Diets, New Brunswick, NJ, USA	D12450
Viscover ExiTron nano 12000	Milteny Biotec, Bergisch Gladbach, Germany	130-095-698
VivoQuant	Invicro
X-CUBE	Molecubes, Belgium		https://www.molecubes.com/systems/