Imaging Non-alcoholic Fatty Liver Disease Model Using H-1 and F-19 MRI

Abstract

Purpose

We explore the use of intravenously delivered perfluorocarbon (PFC) nanoemulsion and ¹⁹F MRI for detecting inflammation in a mouse model of non-alcoholic fatty liver disease (NAFLD). Correlative studies of ¹H-based liver proton density fat fraction (PDFF) and T₁ measurements and histology are also evaluated.

Procedures

C57BL/6 mice were fed standard or high-fat diet (HFD) for 6 weeks to induce NAFLD. ¹H MRI measurements of PDFF and T₁ relaxation time were performed at baseline to assess NAFLD onset prior to administration of a PFC nanoemulsion to enable ¹⁹F MRI of liver PFC uptake. ¹H and ¹⁹F MRI biomarkers were acquired at 2, 21, and 42 days post-PFC to assess changes. Histopathology of liver tissue was performed at experimental endpoint.

Results

Significant increases in liver volume, PDFF, and total PFC uptake were noted in HFD mice compared to Std diet mice. Liver fluorine density and T₁ relaxation time were significantly reduced in HFD mice.

Conclusions

We demonstrated longitudinal quantification of multiple MRI biomarkers of disease in NAFLD mice. The changes in liver PFC uptake in HFD mice were compared with healthy mice that suggests that ¹⁹F MRI may be a viable biomarker of liver pathology.

Introduction

Non-alcoholic fatty liver disease (NAFLD) is a highly prevalent condition affecting approximately 25% of the global population [1]. Untreated NAFLD can progress to non-alcoholic steatohepatitis (NASH) and ultimately result in development of cirrhosis and hepatocellular carcinoma. Ultrasound is a routine imaging approach to clinical diagnosis of NAFLD [2]. However, assessment of ultrasound images for clinical care remains qualitative and therefore subjective. Ultrasound elastography provides a more quantitative approach to assessment [3]; however, this method relies on measurement of tissue stiffness to detect liver fibrosis, which occurs at later disease stages. Overall, invasive procedures including liver biopsy remain the gold standard for assessing fatty liver disease. Quantitative, non-invasive biomarkers that are sensitive to early changes in liver pathology remain a major need for diagnosis and management of patients with NAFLD.

Proton MRI-based biomarkers have been explored to enable liver fat and fibrosis estimation through measurement of proton density fat fraction, tissue relaxivity (T₁ relaxation time), and liver stiffness as measured by MR elastography [4-9]. A direct measure of chronic liver inflammation is not yet available and represents an unmet clinical need as an earlier indicator of NAFLD progression [10]. Chronic inflammation can be detected in tissue specimens as changes in Kupffer cell activation and peripheral leukocyte infiltration prior to widespread microstructural changes associated with fibrosis [11].

To assay the inflammatory component of NAFLD in vivo and noninvasively, we explored whether intravenously delivered perfluorocarbon (PFC) nanoemulsion, followed by quantitative ¹⁹F MRI, can detect changes in liver uptake due to NAFLD onset. Following systemic administration, PFC nanodroplets are phagocytosed by cells of the reticuloendothelial system (RES), particularly macrophages and neutrophils, as part of the clearance pathway for these materials. Accumulation of these in situ labeled cells at inflammatory sites yields “hot-spot” ¹⁹F images that can be quantified accurately. PFC nanoemulsions have been used successfully for imaging inflammation in numerous animal models [12-14]. In normal liver, Kupffer macrophages readily take up PFC nanoemulsions. However, with NAFLD onset, we hypothesize that total PFC liver uptake would be modulated over wild-type levels. In this study, we used a high-fat Western diet mouse model to explore changes in liver PFC accumulation and compared the findings to standard ¹H-based MRI biomarkers of NAFLD and NASH, which develops during late stages of steatosis. The high-fat, Western diet model is characterized by development of obesity and steatosis initially, with eventual progression to fibrosis with prolonged diet administration, and thus resembles human NASH in some respects.

Methods

Animals and Experimental Design

Female B6.Cg-Lep^ob/J mice on C57BL/6 background (8–10 weeks old) from Jackson Laboratories (Bar Harbor, ME) were fed a high-fat diet (HFD, N = 15) containing 40% fat, 2% cholesterol, and 20% fructose (D09100310, Research Diets, New Brunswick, NJ) ad libitum to induce fatty liver disease. Control C57BL/6 mice (N = 6) of the same sex and age were fed a standard (Std) rodent diet containing 13% fat, 0.03% cholesterol, and 0.2% fructose (PicoLab Rodent Diet 5053, LabDiet, St. Louis, MO). After 6 weeks of diet consumption, a single dose of PFC nanoemulsion imaging agent (V-Sense 1000H, Celsense, Pittsburgh, PA) was administered intravenously (0.2 ml containing ~ 1.2 × 10²¹ F atoms). For histology purposes, three animals in each group received a fluorescently labeled version of the PFC nanoemulsion (VS-1000H DM Green) containing a fluorescein isothiocyanate (FITC) fluorophore bound to the PFC droplet, with excitation/emission wavelengths equal to 490/525 nm. The fluorescently labeled PFC exhibits the same distribution and kinetics as the nonfluorescent material. Body weights were taken at least weekly. All animal experiments followed protocols that were approved by University of California San Diego’s Institutional Animal Care and Use Committee (IACUC).

Magnetic Resonance Imaging

Longitudinal in vivo MRI was performed, starting at 6 weeks post-diet initiation, and at 21 and 42 days thereafter, to assess biomarker modulation. Figure 1a shows a schematic of the overall study design. MRI was performed using an 11.7 T small animal imaging system (BioSpec, Bruker, Billerica, MA) equipped with a ¹⁹F/¹H volume coil. Mice were anesthetized for procedures using 1–2% isoflurane in oxygen and warmed using an MRI-compatible air heating system (SA Instruments, Inc., Stony Brook, NY). Respiratory rate and rectal temperature were monitored. A combination of MRI scans was performed to assess biomarkers of liver pathology. ¹⁹F rapid relaxation and enhancement (RARE) images were acquired to assess fluorine concentration in the liver and corresponding ¹H RARE images were acquired for anatomical reference. A reference tube of dilute (10% v/v) PFC was included in the field of view to enable absolute F-atom quantification of image hotspots. Additionally, a multi-gradient echo (MGE) acquisition was used for quantification of liver proton density fat fraction (PDFF) [6, 7, 15] and T₂*. Due to the short T₂* of tissues when working at 11.7 T, multiple MGE sequences were acquired with interleaved echo times to increase the number of sampling points of the echo train with rapid T₂* decay. Also, a ¹H RARE acquisition with multiple repetition time (TR) values were acquired to assess the average liver T₁ relaxation time including fat and water. Detailed parameters for each sequence are outlined in Table 1.

Fig. 1.

Study timeline, weight change, and gross necropsy images. Panel a shows a schematic of the overall study design and number of mice imaged per time point. Panel b displays body weight change over the course of the study from the start of diet administration. Std diet mice show progressive weight gain from the start of diet administration to the time of PFC injection (day 0) which is sustained to the end of the experiment (day 42). HFD mice show significantly increased weight (34%, p < 0.001) compared to Std diet controls on day 0 which is sustained to the end of the study (48%, p = 0.001). In panel c, we display representative photographic images of gross anatomical differences (enlargement, discoloration, tissue mottling) between the livers of Std diet and HFD mice.

Table 1.

Summary of MRI parameters

Imaging protocola	Read-out	TR (ms)	TE (ms)	Echo Spacing (ms)	# Echoes	# Avg	In-plane Res (mm)	Scan time (min)
19F RARE	19F PFC	1000	20	2.5	8	100	0.83 × 0.94	30
1H RARE	Anatomy	1500	17.5	2.19	8	8	0.16 × 0.16	1
MGEb	PDFF	300b	1.47	2.21	8	6	0.23 × 0.23	1
			1.77
			2.23
			2.7
T1 map (RARE)	T1 relax time	646, 1000, 3000, 5500	6.5	3.25	2	2	0.31 × 0.31	1

^aSlice thickness = 2 mm for ¹⁹F RARE, 1 mm for all other protocols

^bFlip angle = 30°, respiratory-gated

Image Data Analysis

Image data were analyzed using VivoQuant 4.0 software (VQ, Invicro, Boston, MA). The ¹⁹F/¹H RARE images were co-registered using a standard affine transformation. The liver region of interest (ROI) was segmented manually using the ¹H image to calculate total liver volume, which was directly mapped to the ¹⁹F data to measure the total ¹⁹F signal intensity. Signal intensity was calibrated to total number of fluorine atoms using a calibrated reference tube in the image, as described elsewhere [13, 16, 17]. The ¹⁹F data were also normalized to the segmented ¹H MRI liver volume to calculate the fluorine content per tissue volume. Using ¹H MGE images, liver volumes were manually segmented from all image slices in which the liver was visible using an image from a single echo. Quantification of PDFF [6, 7, 15, 18] was performed in VQ using a multistep adaptive fitting algorithm that is based on a method previously described by Zhong et al. [18]. T₁ maps were reconstructed using Bruker on-board software. Mean T₁ relaxation times were measured directly from liver ROIs manually defined across all liver slices using VQ without correction to separate fat/water.

Histology

Mice (N = 5) were euthanized at 2 and 42 days post-PFC injection for correlative histology. Samples were either fixed for staining with hematoxylin and eosin (H&E), or fresh-frozen in optimal cutting temperature (OCT) freezing medium for immunohistochemical (IHC) staining. For fixation, mice were perfused transcardially with saline followed by 10% neutral buffered formalin, and excised livers were stored in formalin for 72 h and then transferred to 70% ethanol. Samples for IHC were harvested fresh, rinsed in cold PBS, and then embedded in OCT and frozen at − 80 °C.

The NASH Clinical Research Network (CRN) scoring system [19] was used for evaluation of steatosis and steatohepatitis in mouse livers. Hepatocellular steatosis (scale 0–3), lobular inflammation (scale 0–3), and ballooning (scale 0–2) were scored using the H&E slides by a pathologist blinded to treatment. Anti-mouse F4/80 antibody (Clone 8, BioLegend, San Diego, CA) was used to identify macrophages by IHC. Localization of macrophages with the fluorescein isothiocyanate-labeled PFC nanoemulsion was performed using fluorescent micrographs by an independent histologist blinded to treatment.

Statistical Analysis

All measurements are presented as mean ± standard error of mean (SEM). Results were analyzed for significance using an analysis of variance (ANOVA), with post hoc analysis using Tukey’s honestly significance difference (HSD) test. A p value of < 0.05 was considered statistically significant.

Results

Body Weights and Gross Liver Observations

All HFD mice showed significantly more weight gain (38.7%, p < 0.01) prior to the first imaging time point at 6 weeks compared to Std diet control animals (15.1%, Fig. 1b). At study endpoint (42 days), HFD mice continued to experience significant increases in weight for a net gain of 48.3% (p < 0.01). Upon necropsy, HFD mice displayed visible liver enlargement, corresponding to differences in MRI-derived liver volumes, and pale tissue coloration in all cases compared to Std diet mice (Fig. 1c).

MRI Detects Differences Between HFD and Standard Diet

Our multi-parametric MRI protocol successfully enabled longitudinal quantification of multiple relevant biomarkers including ¹⁹F MRI with a PFC nanoemulsion, PDFF quantification, ¹H T₁ mapping, and ¹H MRI measurements of liver volume, over a 42-day imaging period. Figure 2 displays representative ¹H /¹⁹F and PDFF liver images, and a summary of image quantification results is displayed in Fig. 3. In comparing HFD to the standard diet, striking visual differences were observed in liver volume and PDFF, where both parameters were significantly higher in the HFD animals (Fig. 2).

Fig. 2.

Representative ¹H, ¹⁹F, and PDFF liver images. Panel a displays a ¹H/¹⁹F MRI overlay of PFC uptake in the liver. Std diet control mice show high PFC uptake in the liver at 2 days after PFC administration, followed by PFC clearance over time as expected. HFD mice display clearly reduced PFC density in the liver compared to Std diet controls with a similar clearance trend over time. In b, we display MRI-based ¹H PDFF maps overlaid on anatomical liver images. PDFF maps show high, heterogenous fat signal in HFD mice.

Fig. 3.

MRI biomarker panel. HFD mice display (a) significant liver enlargement compared to Std diet mice (p < 0.01). The total liver PFC content (b) is significantly higher in HFD mice at 42 days post-PFC injection (p < 0.01). When normalizing total liver PFC to tissue volume, liver PFC density (c) is significantly lower in HFD mice than Std diet controls at 21 and 42 days post-treatment (p < 0.01). Proton MRI shows significant increases in PDFF (d) and significant reduction in T₁ relaxation times (e) in HFD mice (p < 0.01). (* denotes statistical significance p < 0.05).

Liver volumes in HFD mice were significantly larger than Std diet mice at all time points (p < 0.01, Fig. 3a), with an overall increase of ~ 12% during the course of the study (42 days). The total liver PFC (¹⁹F) uptake in HFD mice was also significantly higher than Std diet mice (80%, p < 0.0l, Fig. 3b). Std diet mice show an overall decrease from injection to 42 days post, which is consistent with standard clearance of the PFC based on an in vivo half-life of > 100 days [20]. The liver PFC density was lower in the HFD mice, when compared with the Std. diet mice (Fig. 3c). The difference in PFC density between HFD and Std diet mice reached the level of statistical significance at 21 days (41%, p = 0.02) and 42 days (57%, p < 0.01) post-PFC injection (Fig. 3c).

Baseline MRI assessment of HFD mice after 6 weeks on diet displayed statistically significantly higher liver PDFF (660%, p < 0.01) and lower T₁ relaxation time (57%, p < 0.01) compared to Std diet mice (Fig. 3d, e). The PDFF values for HFD animals (Fig. 3d) remained significantly higher than Std diet controls at all time points from 2 days (660%, p < 0.01) to 42 days (600%, p ≤ 0.01) post-PFC injection. Additionally, the ¹H T₁ relaxation times for the HFD cohort (Fig. 3e) were significantly lower from 2 days (52%, p < 0.01) to 42 days (59%, p < 0.01).

Histopathology

Histopathological analyses of liver tissues at experimental endpoints displayed differences in standard and high fat diet. Mice fed Std diet showed no evidence of liver steatosis, lobular inflammation, portal inflammation, necrosis, or ballooning at 2 or 42 days post-PFC injection in H&E-stained sections (Fig. 4). The HFD mice showed significant diffuse steatosis, grade 3, at 2 and 42 days post-PFC injection at both time points (p = 0.03 and 0.008, respectively) in approximately 90% of the liver tissue. No significant portal inflammation, necrosis, and ballooning were observed in any HFD mice, confirming that this disease regimen approximated NAFLD and not NASH.

Fig. 4.

Liver histology. Representative images of H&E staining of liver tissues (top row) show significant steatosis and lobular inflammation (both mean scores = 3) in HFD mice compared to Std diet mice (both scores = 0) at 42 days after PFC injection. Confocal microscopy of representative IHC slides (bottom row) confirms increased presence of F4/80 + macrophage staining in HFD mice compared to Std diet controls.

Immunohistochemistry of liver cryosections showed the presence of cells of macrophage phenotype (i.e., Kupffer cells and peripheral macrophages) via F4/80 antibody staining (Fig. 4), along with sparse and diffuse fluorescent-PFC accumulation in the livers of Std diet and HFD mice. The sparsity of Kupffer cell labeling is consistent with prior studies labeling macrophages in various inflammation models [12-14, 21]. Increased macrophage presence was grossly observed in HFD mice (Fig. 4).

Discussion

The MRI-based in vivo biomarkers evaluated in this study detected the impact of HFD on liver. Liver volume and PDFF were particularly sensitive to the disease as expected. The large liver volumes observed in the HFD mice is generally associated with lipid droplet formation. These findings are consistent with significant increases in steatosis as confirmed by histological results. Additionally, the total F uptake and density decreased with disease progression. The total F uptake is sensitive to liver volume (Fig. 3b). However, the F density (Fig. 3c) may be directly sensitive to steatosis, which presumably reduces the density of viable Kupffer cells that take up PFC. Alternatively, viable Kupffer cells could be saturated with lipid droplets thereby reducing the capacity to take up PFC droplets. Moreover, the bulk in vivo F-atom measurements (Fig. 3b) do not distinguish between resident Kupffer cells and peripheral macrophages that may enter the liver as a consequence of inflammation during disease pathogenesis; these two cell populations predominately harbor the PFC agent. Moreover, the degree of potential Kupffer cell activation, which may affect PFC uptake, or change in Kupffer cell density during disease course, is unknown. Future experiments involving flow cytometry of disassociated liver tissues may further clarify these questions by separately quantifying and phenotyping Kupffer and peripheral cell populations that are labeled with fluorescent PFC nanoemulsion, but these experiments are beyond the scope of the study. Generally, in long-term longitudinal studies, if a labeled macrophage dies, it can transfer the label to another macrophage that engulfs the dying cell body, thereby resulting in a new labeled macrophage; alternatively, released PFC droplets from an apoptotic macrophage could be redistributed to Kupffer cells. Detailed histological studies have observed macrophage labeling > 30 days post-agent injection [12]. Also, gradual clearance of the PFC agent in liver is observed in these studies (Fig. 3b, c); to refine longitudinal measurements, of one could apply an exponential clearance correction to account for slow clearance.

In summary, chronic administration of HFD resulted in a marked increase in liver volume and steatosis of all animals indicative of NAFLD. The PFC nanoemulsion imaging agent enabled detection of liver disease progression and may provide additional information to more conventional PDFF and T₁ biomarkers. In future work, initiation of imaging earlier in the disease time course could enable improved detection of the early disease onset, especially early the inflammation component via PFC quantification, which could complement biomarkers that are more directly sensitive to steatosis at later stages of NAFLD.

Data Availability

Data that support the findings of this study are available from the corresponding author upon reasonable request.