Dual-input dual-compartmental uptake and efflux model based on Gd-EOB-DTPA-enhanced MRI for simultaneously assessing liver function and fibrosis

Abstract

Background

Liver function assessment and fibrosis staging are crucial for monitoring therapeutic efficacy and guiding surgical management in patients with chronic liver disease. This study aimed to determine whether pharmacokinetic parameters derived from dual-input dual-compartmental uptake and efflux model based on Gd-EOB-DTPA-enhanced magnetic resonance imaging (MRI) could simultaneously assess liver function and fibrosis.

Methods

Thirty rats were enrolled in this study. Thioacetamide (TAA) was administered for 2, 4, 6, and 8 weeks to induce liver fibrosis. The METAVIR system (F0–F4) was used to stage fibrosis. Pharmacokinetic modeling was performed in a voxel-wise manner to generate parametric maps, from which mean values were extracted using liver regions of interest (ROIs). Pharmacokinetic parameters, including plasma flow rate (F_p), extracellular space (v_ecs), arterial supply fraction (f_a), mean uptake (k_i) and efflux (k_ef) rate of hepatocytes, were compared across fibrosis groups. The expression of hepatic transporters, including the organic anion-transporting polypeptide 1a1 (Oatp1a1) and the multidrug resistance-associated protein 2 (Mrp2), served as a reference for liver function. The relationship between pharmacokinetic parameters and hepatic transporters expression was assessed. Receiver operating characteristic (ROC) curve analysis was conducted to assess the diagnostic performance of pharmacokinetic parameters in staging fibrosis.

Results

There were 6, 11, and 13 rats designated into the control (F0), early fibrosis (F1–2), and advanced fibrosis groups (F3–4), respectively. With the progression of liver fibrosis, F_p and k_i decreased significantly, while v_ecs and f_a increased significantly. Compared with the control group, the expression of Oatp1a1 and Mrp2 decreased in rats with liver fibrosis. Significant correlations were observed between F_p, v_ecs, k_i, f_a and Oatp1a1 expression (r=0.877, −0.762, 0.722, −0.460; P<0.05 for all); k_i and f_a were also significantly correlated with Mrp2 expression (r=0.435, P=0.02 and r=−0.475, P=0.008). Further multiple linear regression analysis identified k_i as the only parameter significantly associated with Oatp1a1 expression (beta =0.474; P=0.002), while no parameters were significantly related to Mrp2 expression (P>0.05 for all). For detecting fibrosis (F0 vs. F1–4), the areas under the curve (AUCs) of F_p, v_ecs, and k_i were 0.903, 0.917, and 1.000, respectively. For distinguishing advanced fibrosis (F0–2 vs. F3–4), the AUCs of F_p, v_ecs, and k_i were 1.000, 0.916, and 0.862, respectively.

Conclusions

The pharmacokinetic parameters derived from dual-input dual-compartmental uptake and efflux model, especially for k_i, showed potential for simultaneously assessing liver function and staging fibrosis.

Highlight box.

Key findings

• Pharmacokinetic parameters derived from the dual-input dual-compartmental uptake and efflux model based on Gd-EOB-DTPA-enhanced magnetic resonance imaging (MRI) correlated significantly with hepatic transporter expression and demonstrated excellent diagnostic performance for staging liver fibrosis in a rat model.

What is known and what is new?

• Conventional methods for liver function assessment (e.g., Child-Pugh, albumin-bilirubin, indocyanine green clearance) provide indirect or global measures and lack the ability to evaluate regional hepatic function or fibrosis simultaneously.

• This study shows that quantitative MRI-derived pharmacokinetic parameters (plasma flow rate, extracellular space, the mean uptake rate of hepatocytes, the arterial supply fraction) can reflect both liver perfusion and hepatocyte transport function, enabling simultaneous evaluation of liver function and fibrosis staging, with strong correlations to histopathology and molecular biomarkers.

What is the implication, and what should change now?

• This pharmacokinetic model provides a noninvasive imaging biomarker that may facilitate early detection of fibrosis, comprehensive liver function evaluation, and more accurate preoperative risk stratification in patients with chronic liver disease. Future translational and clinical studies should integrate this pharmacokinetic MRI approach into liver disease assessment protocols. Its adoption could improve monitoring of therapeutic response and surgical planning by moving beyond traditional serological or global measures toward precise, segmental, and functionally relevant imaging-based evaluation.

Introduction

The preoperative evaluation of total and regional liver function is crucial for preventing postoperative liver failure and predicting prognosis in patients with chronic liver disease (1). Currently, quantitative assessment of liver function remains challenging and primarily relies on clinical scoring systems, such as the Child-Pugh score and the albumin-bilirubin (ALBI) score. However, the Child-Pugh score incorporates some subjective clinical assessments, including ascites and hepatic encephalopathy. The ALBI score was much more objective, but it merely provided surrogate parameters for the extent of liver damage (2,3). In contrast, the indocyanine green (ICG) clearance test provides a direct method for assessing liver function, but it can be affected by hyperbilirubinemia, blood oxygen concentration, and other competitive agents (4). Moreover, while these serological methods can assess global liver function, they are unable to evaluate the function of specific liver segments.

Dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) with Gd-EOB-DTPA, a hepatocyte-specific contrast agent, has been proposed to evaluate liver function (5,6). Given that anionic xenobiotics and endogenous substances are metabolized and excreted through the organic anion transporter system, the in vivo expression of organic anion transporters is an important indicator of hepatic function (7). With the progression of chronic liver disease, such as liver fibrosis, the expression of organic anion transporters decreases, contributing to a decline in liver function and consequently reducing Gd-EOB-DTPA uptake (8,9).

Previous studies measured the relative enhancement (RE) between the plain and contrast-enhanced signal intensity (SI) of liver parenchyma or the reduction rate of T1 relaxation time (rrT1) to explore the value of Gd-EOB-DTPA in evaluating liver function (5,6). However, some of these studies lacked pathological results as a reference, and T1 values of tissues could be affected by field strength (5). Additionally, these studies did not establish an effective model to describe the pathophysiological changes in the progression of chronic liver disease.

Quantitative assessment of liver function based on pharmacokinetic models of Gd-EOB-DTPA has become a research focus in recent years. The imaging parameters calculated from these pharmacokinetic models not only reflect the changes in hepatic contrast concentration over time but also provide insights into blood perfusion in the liver (10). Based on the pharmacokinetic properties of Gd-EOB-DTPA, Georgiou (11) proposed the dual-input dual-compartmental uptake and efflux model to estimate the hepatic uptake and efflux transport processes of gadoxetate in healthy volunteers in vivo. The results suggested that parameters derived from this pharmacokinetic model could quantitatively reflect liver perfusion as well as the mean uptake and efflux rates of hepatocytes. However, the value of this pharmacokinetic model in assessing liver function and monitoring liver pathology remains to be further investigated.

Therefore, we aimed to establish a liver fibrosis model in rats and investigate whether the proposed dual-input dual-compartmental uptake and efflux model (11) could quantitatively assess liver function and stage liver fibrosis, using histopathological results as a reference. This article is presented in accordance with the ARRIVE reporting checklist (available at https://tgh.amegroups.com/article/view/10.21037/tgh-25-135/rc).

Methods

Animal model

Experiments were performed under a project license (approved No. 20223762) granted by the Ethics Committee of Tongji Medical College, Huazhong University of Science and Technology, in compliance with the institutional guidelines for the care and use of animals. A protocol was prepared before the study without registration.

Thirty male Sprague-Dawley rats (8 weeks old, 200±20 g) were randomly divided into control (n=6) and fibrosis groups (n=24). Liver fibrosis was induced by intraperitoneal injection of thioacetamide (TAA; Sigma-Aldrich, Spain) dissolved in normal saline three times a week at a dose of 250 mg/kg for 2, 4, 6, and 8 weeks (Figure 1A) (12). Control group rats received intraperitoneal injection of normal saline at the same dose and frequency for 8 weeks. All rats were maintained under humane conditions with adequate food and water.

Figure 1.

Flowchart of this study. (A) The experimental schedules of the fibrosis group and the control group. (B) The dual-input dual-compartmental uptake and efflux model used in the DCE-MRI scanning. DCE-MRI, dynamic contrast-enhanced magnetic resonance imaging; f_a, the arterial supply fraction; F_p, plasma flow rate; k_ef, the mean efflux rate of hepatocytes; k_i, the mean uptake rate of hepatocytes; RT-PCR, reverse transcription polymerase chain reaction; TAA, thioacetamide; v_ecs, extracellular space.

MRI acquisition

The imaging procedure was carried out using a 3.0 T MRI device (Ingenia 3.0 T, Philips Healthcare, Best, the Netherlands) with an eight-channel phased-array rat coil (70-mm diameter; Shanghai Chenguang Medical Technologies, China).

MR imaging was performed 2 days after the last TAA injection, which allowed time to minimize acute inflammation in the liver parenchyma. Prior to scanning, the rats were anesthetized with 3% pentobarbital (w/v; 0.2 mL/100 g body weight) via intraperitoneal injection. A catheter was then placed in the lateral tail vein of each rat. The rats were positioned supine with their heads aligned straight forward. To minimize respiratory motion, their abdomens were secured with a belt.

DCE-MRI data were acquired in the axial plane using the single-shot spoiled gradient recalled echo T1-weighted sequence. The detailed imaging parameters were as follows: field of view (FOV), 70 mm × 70 mm; voxel size, 0.5×0.5×2 mm; matrix, 140×140; repetition time/echo time (TR/TE), 9.6 ms/2.7 ms; flip angle, 15 degrees; interslice gap, 5 mm. A total of 35 dynamic phases were obtained. The temporal resolution per dynamic phase was approximately 80–90 seconds, determined by the TR, matrix size, number of slices, and single-shot acquisition. At the beginning of the fifth dynamic phase, a bolus of 0.025 mmol/kg Gd-EOB-DTPA followed by a 1-mL saline flush was manually injected through the tail vein, which was completed within 3–5 seconds. The contrast injection was timed to the fifth phase rather than the start of the acquisition to allow collection of several stable pre-contrast baseline phases, which are essential for accurate SI-to-concentration conversion, correction of system drift, and robust voxel-wise pharmacokinetic fitting. The total duration of the dynamic acquisition was therefore approximately 49–50 minutes (35 phases × ~85 seconds per phase) (11).

Image processing and pharmacokinetic modeling

The dynamic data were preprocessed using the FMRIB Software Library (FSL) package (13), including rigid body registration for motion correction and basic image algebra to calculate the contrast agent concentration at each dynamic time point. To generate venous and arterial input function signal-intensity time curves, regions of interest (ROIs) were manually placed in the main portal vein (at the porta hepatis level) and the proximal abdominal aorta (at the celiac axis level) on a single slice at one time point. These ROIs were subsequently propagated throughout the dynamic series.

In DCE-MRI, the relationship between SI and Gd-EOB-DTPA concentration was considered to be linear (14), and the contrast agent concentration was calculated according to the following formula:

$$\begin{array}{l}C\left(t\right)=k\left[S\left(t\right)-S_0\right]/S_0\hfill \end{array}$$ [1]

In this formula, C(t) is the tissue tracer concentration at time t, S₀ is the unenhanced SI, S(t) is the contrast-enhanced SI at time t, and k is the scaling constant (0.395 for the liver and 0.201 for blood). These constants are derived from a previous study (15).

We then applied the dual-input dual-compartmental uptake and efflux model, as proposed by Georgiou et al. (11), to describe the pharmacokinetic properties of Gd-EOB-DTPA in the liver. The model is defined in Eq. [2] and illustrated in Figure 1B. A detailed introduction to Eq. [2] can be found in the previous study (11).

$$\begin{array}{l}{C}_l\left(t\right)=F_p·\left\{\frac{E_i}{1-T_e/T_i}{e}^{-t/T_i}+\left(1-\frac{E_i}{1-T_e/T_i}\right)e^{-t/T_e}\right\}\ast c_p\left(t\right)\\ C_p\left(t\right)=\left(f_a{c}_a\left(t\right)+f_v{c}_v\left(t\right)\right)/\left(1-Hct\right)\end{array}$$ [2]

The conceptual framework of the model is based on the liver parenchyma being composed of two main components: (I) the extracellular space (v_ecs), which is further divided into the vascular space (i.e., plasma) and the interstitial space; and (II) the intracellular space (v_i), which primarily consists of hepatocytes. The contrast agent gains access to the liver parenchyma via a dual-route mechanism. The weighted sum of the arterial supply fraction (f_a, corresponding to the hepatic artery) and the venous supply fraction (f_v, associated with the hepatic portal vein) determines the relative contributions of each pathway to the overall input. It is postulated that the transfer of the contrast agent between the plasma and interstitial spaces occurs at a very fast rate (i.e., with endothelial permeability being extremely high, approaching infinity), leading to an almost instantaneous state of equilibrium.

In this study, pharmacokinetic modeling was performed in a voxel-wise manner. After motion correction and conversion of SI to contrast concentration, the dual-input dual-compartmental uptake and efflux model was fitted for each voxel within the liver to generate parametric maps. The main parameters were as follows: (I) plasma flow rate (F_p), which reflects the total liver blood flow; (II) extracellular space (v_ecs); (III) arterial supply fraction (f_a); (IV) the mean uptake (k_i); and (V) efflux (k_ef) rate of hepatocytes. Three ROIs were manually placed in the liver parenchyma on the central slice of each parametric map to extract the mean values from all included voxels, taking care to avoid artifacts, major vessels, bile ducts, and liver margins. ROI placement and pharmacokinetic parameter calculation was performed by two radiologists (T.G. and Y.G., with 10 and 5 years of experience in abdominal imaging, respectively) who were blinded to the histopathological results so that we could calculate intraclass correlation coefficients (ICCs).

Histopathology

After each MRI scan, the rats were humanely sacrificed at 2, 4, 6, and 8 weeks, and their livers were excised (Figure 1A). The liver samples were preserved in 10% phosphate-buffered formalin. Some sections underwent hematoxylin-eosin (H&E) staining for qualitative morphological evaluation of the liver tissue, while other fixed liver tissues were stained with Sirius red and subjected to α-smooth muscle actin (α-SMA) immunohistochemical staining to determine the extent of fibrosis (16). Sirius Red-positive areas, indicating collagen deposition, and α-SMA-positive areas, reflecting activated hepatic stellate cells, were quantified using ImageJ software (version 1.52a; National Institutes of Health). The results were expressed as the percentage of positively stained area relative to the total tissue area (%), with higher values representing greater fibrosis severity and lower values indicating milder fibrosis. A pathologist with over ten years of experience in liver pathology reviewed all pathological specimens. Fibrosis stages were assessed based on the METAVIR system (17), defined as: F0 (no fibrosis), F1 (portal fibrosis without septa), F2 (portal fibrosis with a few septa), F3 (numerous septa without cirrhosis), and F4 (cirrhosis). In the subsequent analysis, F0 served as the control group, F1 and F2 as the early fibrosis group, and F3 and F4 as the advanced fibrosis group.

Real-time reverse transcription polymerase chain reaction (real-time RT-PCR)

Additional liver samples were collected, snap-frozen in liquid nitrogen, and stored at −80 ℃ for subsequent real-time RT-PCR analysis. Studies have demonstrated that gadoxetate is transported into rat hepatocytes via organic anion-transporting polypeptide 1a1 (Oatp1a1) and is excreted into the bile through multidrug resistance-associated protein 2 (Mrp2) (10,18). RNA extraction, cDNA synthesis, and real-time RT-PCR procedures were performed as described in a previous study (19). The primers for Oatp1a1 and Mrp2 were listed in Table S1. The expression levels of these transporters were normalized to those of the housekeeping gene, β-glucuronidase (GUSB), which has been validated as a stable reference gene under our experimental conditions (20). Relative mRNA expression was calculated using the 2^−ΔΔCt method. Higher 2^−ΔΔCt values represent higher transporter expression, whereas lower values indicate reduced expression.

Statistical analysis

SPSS 25.0 (Chicago, IL, USA) and GraphPad Prism (Version 8.0.1, San Diego, CA, USA) were used to perform all statistical analyses. Quantitative data with a normal distribution were expressed as the mean ± standard deviation, while non-normally distributed data were presented as the median with interquartile range. Interobserver reproducibility of pharmacokinetic parameters was evaluated using the ICC, calculated with a two-way random-effects model and absolute-agreement definition to assess measurement consistency between the two observers. Pharmacokinetic parameters with an ICC >0.75 were included, and the data measured by the more experienced reviewer were used for subsequent analysis. Statistical differences in pharmacokinetic parameters, histopathological metrics, and the expression of hepatic transporters among different groups were evaluated using one-way analysis of variance (ANOVA) with post-hoc least significant difference (LSD) test or the Kruskal-Wallis test. Spearman correlation analysis was conducted to assess the relationship between these variables. The diagnostic performance of pharmacokinetic parameters in staging liver fibrosis was determined using receiver operating characteristic (ROC) curve analysis and Delong test. A P value <0.05 was regarded as statistically significant.

Results

Histopathological analysis and liver fibrosis staging

H&E staining demonstrated that as liver fibrosis advanced, the organization of hepatocyte cords became increasingly disrupted (Figure 2). Additionally, collagen fiber deposition increased with the development of liver fibrosis, as Table 1 indicated that the percentages of positive areas for Sirius red and α-SMA staining rose significantly with increasing fibrosis severity (P<0.001 for both). Based on the histopathological findings, the rats were assigned to the control (n=6), early fibrosis (n=11), and advanced fibrosis (n=13) groups, respectively.

Figure 2.

Representative histopathological examples of different liver fibrosis stages. (A) H&E staining; (B) Sirius red staining; (C) α-SMA staining. H&E, hematoxylin-eosin; α-SMA, α-smooth muscle actin.

Table 1. Pharmacokinetic parameters and histopathological metrics of different liver fibrosis stages.

Variable	Control	Early fibrosis	Advanced fibrosis	P value
Fp (mL/min/mL)	2.93±0.05	2.81±0.12	2.26±0.19*†	<0.001
vecs (mL/mL)	0.40±0.07	0.49±0.06*	0.57±0.05*†	<0.001
ki (/min)	0.07±0.01	0.03±0.01*	0.02±0.01*†	<0.001
kef (/min)	0.02±0.01	0.03±0.01	0.03±0.01	0.08
f a	0.48±0.15	0.59±0.18	0.68±0.13*	0.03
Sirius red-positive ratio (%)	2.10 (1.60, 2.30)	5.18 (5.06, 7.16)	14.99 (14.33, 19.42)*†	<0.001
α-SMA-positive ratio (%)	1.47 (1.29, 1.77)	4.33 (3.85, 6.39)	13.80 (11.23,16.05)*†	<0.001

Quantitative data with a normal distribution were expressed as the mean ± standard deviation, while non-normally distributed data were presented as the median with interquartile range. *, P<0.05 vs. control; ^†, P<0.05 vs. early fibrosis. f_a, the arterial supply fraction; F_p, plasma flow rate; k_ef, the mean efflux rate of hepatocytes; k_i, the mean uptake rate of hepatocytes; v_ecs, extracellular space; α-SMA, α-smooth muscle actin.

Interobserver reproducibility of pharmacokinetic parameters

The interobserver reproducibility of pharmacokinetic parameters was excellent, with the following ICCs for the two radiologists: F_p, 0.861 [95% confidence interval (CI): 0.728–0.931]; v_ecs, 0.853 (0.714–0.927); k_i, 0.867 (0.739–0.935); k_ef, 0.789 (0.602–0.894); and f_a, 0.776 (0.583–0.887).

Changes of pharmacokinetic parameters in different fibrosis grades

All pharmacokinetic parameters followed a normal distribution. As shown in Table 1, F_p and k_i decreased, while v_ecs and f_a increased with the progression of liver fibrosis (P<0.001 for all). Except for k_ef (P=0.08), other parameters such as F_p, v_ecs, k_i, and f_a showed statistically significant differences among the fibrosis groups (P<0.05 for all). Thus, k_ef was excluded from subsequent analyses. Representative maps of the pharmacokinetic parameters were shown in the Figure 3.

Figure 3.

Typical maps of the pharmacokinetic parameters, including the F_p, v_ecs, k_i, and f_a, with different fibrosis stages. f_a, the arterial supply fraction; F_p, plasma flow rate; k_i, the mean uptake rate of hepatocytes; v_ecs, extracellular space.

Expression of hepatic transporters and its correlation with pharmacokinetic parameters

Real-time RT-PCR analysis revealed that the expression of Oatp1a1 and Mrp2 decreased as liver fibrosis progressed. The expression of Oatp1a1 was 1.96 (1.60, 2.03) in the control group, 1.15 (1.07, 1.32) in the early fibrosis group, and 0.81 (0.75, 0.87) in the advanced fibrosis group. Significant differences were observed among these three groups (P<0.001) (Figure 4A). In contrast, the expression of Mrp2 was 1.23 (1.13, 1.66) in the control group, 1.10 (0.81, 1.63) in the early fibrosis group, and 0.96 (0.64, 1.28) in the advanced fibrosis group, with no significant differences across the groups (Figure 4B).

Figure 4.

Relative expression comparison of (A) Oatp1a1 and (B) Mrp2 in the control group (n=6), early fibrosis group (n=11), and advanced fibrosis group (n=13). ***, P<0.001; ns (not significant), P>0.05.

Significant correlations were found between Oatp1a1 expression and the pharmacokinetic parameters (r=0.877 for F_p, −0.762 for v_ecs, 0.722 for k_i, and −0.460 for f_a; P<0.05 for all). Correlations between Mrp2 expression and pharmacokinetic parameters were also observed (r=0.435 for k_i, P=0.02; r=−0.475 for f_a, P=0.008) (Figure 5). Further multiple linear regression analysis identified k_i as the only parameter significantly associated with Oatp1a1 expression (beta =0.474; P=0.002), while no parameters were significantly related to Mrp2 expression (P>0.05 for all).

Figure 5.

Spearman correlations between the expression of hepatic transporters and the pharmacokinetic parameters. (A) Correlation between the expression of Oatp1a1 and the F_p, v_ecs, k_i, and f_a. (B) Correlation between the expression of Mrp2 and F_p, v_ecs, k_i, and f_a. f_a, the arterial supply fraction; F_p, plasma flow rate; k_i, the mean uptake rate of hepatocytes; v_ecs, extracellular space.

Correlations between pharmacokinetic parameters and histopathological metrics

As illustrated in Figure 6, spearman correlation coefficients between F_p, v_ecs, and k_i and Sirius red-positive ratios were −0.875, 0.833, and −0.740, respectively (P<0.001 for all). Correlation coefficients for F_p, v_ecs, and k_i relative to α-SMA-positive ratios were −0.837, 0.812, and −0.725, respectively (P<0.001 for all). f_a exhibited moderate correlations with histopathological metrics (r=0.465 and 0.485 for Sirius red and α-SMA-positive ratios, respectively; P<0.05 for both).

Figure 6.

Spearman correlations between the histopathological metrics and the pharmacokinetic parameters. (A) Correlation between the Sirius red-positive ratio and the F_p, v_ecs, k_i, and f_a. (B) Correlation between the α-SMA-positive ratio and F_p, v_ecs, k_i, and f_a. f_a, the arterial supply fraction; F_p, plasma flow rate; k_i, the mean uptake rate of hepatocytes; v_ecs, extracellular space; α-SMA, α-smooth muscle actin.

Diagnostic performance of pharmacokinetic parameters for staging liver fibrosis

To determine the presence of fibrosis (F0 vs. F1–4), ROC curve analyses revealed that the area under the curve (AUC) values for F_p, v_ecs, k_i, and f_a were 0.903, 0.917, 1.000, and 0.792, respectively (Table 2; Figure 7A). However, Delong test suggested that no significant difference was found among the ROC curves for these parameters (P>0.05).

Table 2. Diagnostic efficiency of pharmacokinetic parameters in differentiating liver fibrosis stages.

Liver fibrosis stage	Parameter	AUC (95% CI)	Sensitivity (%)	Specificity (%)	P value
F0 vs. F1–4	F p	0.903 (0.738–0.980)	87.5	100	<0.001
	v ecs	0.917 (0.757–0.986)	87.5	100	<0.001
	k i	1.000 (0.884–1.000)	100	100	<0.001
	f a	0.792 (0.605–0.917)	79.2	83.3	0.01
F0–2 vs. F3–4	F p	1.000 (0.884–1.000)	100	100	<0.001
	v ecs	0.916 (0.756–0.986)	92.3	88.2	<0.001
	k i	0.862 (0.687–0.960)	100	70.6	<0.001
	f a	0.706 (0.512–0.857)	100	47.1	0.04

AUC, area under the curve; CI, confidence interval; f_a, the arterial supply fraction; F_p, plasma flow rate; k_i, the mean uptake rate of hepatocytes; v_ecs, extracellular space.

Figure 7.

ROC curves for differentiation of liver fibrosis stages with F_p, v_ecs, k_i, and f_a. (A) F0 vs. F1–4; (B) F0–2 vs. F3–4. AUC, area under the curve; f_a, the arterial supply fraction; F_p, plasma flow rate; k_i, the mean uptake rate of hepatocytes; ROC, receiver operating characteristic; v_ecs, extracellular space.

For distinguishing advanced fibrosis (F0–2 vs. F3–4), the AUC values for F_p, v_ecs, k_i, and f_a were 1.000, 0.916, 0.862, and 0.706, respectively (Table 2; Figure 7B). F_p and v_ecs exhibited excellent diagnostic performance. No significant difference was found between the ROC curves for F_p and v_ecs (Z=1.625, P=0.10).

Discussion

Our study demonstrated that the pharmacokinetic parameters derived from dual-input dual-compartmental uptake and efflux model were significantly correlated with the expression of hepatic transporters (Oatp1a1 and Mrp2) and presented satisfactory diagnostic efficiency for staging liver fibrosis in a rat model. These findings suggested that the pharmacokinetic model had the potential to simultaneously assess liver function and stage liver fibrosis.

With the progression of liver fibrosis, F_p and k_i decreased, while v_ecs and f_a increased. As liver fibrosis progresses, factors such as hepatocyte degeneration, sinusoid capillarization, and the excessive accumulation of extracellular matrix proteins contribute to the narrowing of the hepatic intracellular space and impact the hemodynamics of hepatic sinusoids (21). The portal vein is the primary source of blood supply to the liver. In contrast to the hepatic artery, the portal vein is more susceptible to compression by adjacent tissues. As a result, the reduction in portal vein perfusion becomes more pronounced, leading to a decrease in overall liver perfusion (22). However, due to the liver’s unique regulatory mechanisms, reduced portal vein perfusion triggers an increase in hepatic artery perfusion as a compensatory response, thereby raising the blood supply fraction from the hepatic artery (23). These characteristics explain the observed changes in F_p, v_ecs, and f_a during the progression of liver fibrosis in this study.

Our study revealed that the expression of Oatp1a1 and Mrp2 gradually decreased with the progression of liver fibrosis, indicating a decline in liver function. This result was consistent with the study by Giraudeau et al. (24). In contrast to Oatp1a1, the expression of Mrp2 showed no significant differences across the three groups. This result was similar to the study by Chen et al. (25). In our study, the mean efflux rate of hepatocytes (k_ef) also did not significantly differ among these groups. k_ef was an imaging biomarker reflecting the transport of Gd-EOB-DTPA to the bile duct system via Mrp2. The results of Mrp2 expression exactly explained the absence of inter-group statistical difference for k_ef from the histopathological perspective. However, in the study by Sheng et al. (26), although the relative expression level of Mrp2 also gradually decreased with the progression of liver fibrosis in rats, significant differences were observed among various fibrosis stages. The reasons for this discrepancy may include: (I) the small sample size in our study, which could have introduced statistical bias; (II) the use of different housekeeping genes in the studies. In our study, we selected β-glucosidase (GUSB) as the housekeeping gene, as a previous study (20) suggested GUSB was more suitable for liver disease research, while Sheng et al. (26) used 3-glyceraldehyde phosphate dehydrogenase (GAPDH). In our study, the observed potential relationship between Mrp2 and k_ef may provide important guidance for optimizing DCE-MRI for clinical use. Because k_ef primarily reflects the hepatocyte excretory phase mediated by Mrp2, its estimation relies less on the early arterial and portal venous dynamics and more on the later portion of the enhancement curve. This implies that a full 40–50-minute continuous dynamic acquisition, as used in our animal protocol, may not be necessary in clinical practice. Instead, a shortened protocol that reduces the number of early dynamic phases while acquiring a limited set of strategically timed delayed images could still provide robust sensitivity to hepatobiliary efflux. Such a design would substantially improve efficiency and feasibility for routine scanning while preserving quantitative accuracy of k_ef. Future work should explore optimized temporal sampling strategies and model-based reconstruction approaches to further reduce scan time without compromising the reliability of hepatobiliary functional assessment.

We further analyzed the correlations between pharmacokinetic parameters and the relative expression of Oatp1a1 and Mrp2. We found that F_p, v_ecs, k_i, and f_a were significantly associated with Oatp1a1 expression; k_i and f_a were also significantly associated with Mrp2 expression (P<0.05). These results indicated that the parameters derived from this pharmacokinetic model were promising for evaluating liver function. Nevertheless, in the multiple linear regression analysis, k_i was found to be the only parameter with a significant correlation to Oatp1a1 expression, and no parameters were significantly associated with Mrp2 expression. Compared with F_p, f_a, and v_ecs, which reflect blood perfusion and liver tissue microstructure, k_i was more directly involved in characterizing hepatocyte ability to transport Gd-EOB-DTPA via Oatp1a1 and was less likely to be influenced by cardiac output or central blood volume. This result suggested that k_i was a useful biomarker for assessing hepatocytic transport function.

For determining fibrosis presence or distinguishing advanced liver fibrosis, F_p, v_ecs, and k_i all demonstrated high AUC values, sensitivity, and specificity. In addition, there were strong correlations between these pharmacokinetic parameters and histopathological metrics (Sirius red and α-SMA-positive ratios), particularly for F_p and v_ecs. As mentioned earlier, F_p and v_ecs reflected blood perfusion and liver tissue microstructure, respectively, while k_i primarily reflected hepatocytic transport function. Previous studies have suggested that liver stiffness measured by ultrasound elastography (USE) or magnetic resonance elastography (MRE) is an important biomarker for staging liver fibrosis (27,28). However, cellular changes precede tissue changes, which in turn precede changes in the organism (29). Thus, the pharmacokinetic parameters derived from the dual-input dual-compartmental uptake and efflux model were also valuable for early detection of liver fibrosis and identification of advanced liver fibrosis.

This study had several limitations. First, the small sample size might have introduced selection bias. Second, the study did not quantitatively evaluate confounding factors such as inflammation, steatosis, and iron deposition. Further studies are necessary to address the impacts of these factors on liver function evaluation and liver fibrosis development. Finally, in clinical practice, considering the heterogeneity of liver function distribution in patients with chronic liver diseases, regional liver function should also be considered when planning for residual liver volume. However, this experimental study did not further analyze the liver function of each hepatic segment, since the liver volume of rats was too small to analyze. Future studies are needed to verify the efficiency of this pharmacokinetic model in assessing segmental liver function in clinical settings.

Conclusions

In our study, we found that k_i derived from the dual-input dual-compartmental uptake and efflux model was a promising biomarker for quantifying hepatocyte transport function and was also useful for staging fibrosis. Since it is an experimental study performed on rats, future studies are warranted to evaluate the efficacy of this pharmacokinetic model for assessing liver function and staging fibrosis in patients with chronic liver disease, which might be helpful for monitoring therapeutic efficacy and guiding surgical managements.

Supplementary

The article’s supplementary files as

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. Experiments were performed under a project license (approved No. 20223762) granted by the Ethics Committee of Tongji Medical College, Huazhong University of Science and Technology, in compliance with the institutional guidelines for the care and use of animals.

Data Sharing Statement

Available at https://tgh.amegroups.com/article/view/10.21037/tgh-25-135/dss

tgh-11-25-135-dss.pdf ^{(68.8KB, pdf)}

DOI: 10.21037/tgh-25-135