Notice: Wiley Online Library will be unavailable on Saturday 27th February from 09:00-14:00 GMT / 04:00-09:00 EST / 17:00-22:00 SGT for essential maintenance. Apologies for the inconvenience.
Potential conflict of interest: Nothing to report.
Liver diseases and regeneration are associated with hemodynamic changes denoting pathological alterations. Determining and monitoring physiological and pathological liver changes is essential for diagnostic and therapeutic objectives. Our aim was to determine the feasibility of functional magnetic resonance imaging (fMRI) during hypercapnia and hyperoxia for monitoring liver pathology. Liver fMRI images were acquired in rodents following acute bleeding, partial hepatectomy, and fibrosis. Results were quantitated and confirmed by histology. Changes induced by hyperoxia and hypercapnia following hemorrhage significantly correlated with the percentage of blood loss, reflecting lower liver perfusion and diminished vessel responsiveness to gas saturation. Hepatectomy resulted in an early decline in signal intensity changes due to hyperoxia, suggesting a decrease in liver perfusion and blood content. Following hepatectomy, signal intensity changes due to hypercapnia increased, signifying a change in liver perfusion from a mainly portal to a more arterial source. Two weeks after induction of fibrosis, signal intensity changes due to hypercapnia became much lower and those due to hyperoxia were much higher than those in normal livers, reflecting the increased perfusion due to the inflammatory process as confirmed by histologic analysis. With fibrosis progression, signal intensity changes induced by hypercapnia and hyperoxia were gradually attenuated, indicating structural and functional alterations of the liver vasculature during fibrosis. Conclusion: In various liver pathologies, fMRI response to hypercapnia and hyperoxia is sensitive to changes in liver hemodynamic status involved in hepatic damage or recovery; thus, this technique may offer an additional noninvasive diagnostic tool for evaluation and follow-up of liver diseases by means of examining perfusion-related alterations. (HEPATOLOGY 2008.)
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Chronic liver diseases are the 12th leading cause of death in the United States and represents 1.1% of all the reported deaths.1, 2 Evaluation of the level of liver pathological changes, such as cirrhosis, regeneration, or cancer, is important from both the diagnostic and the therapeutic aspects. Currently, there are limited tools for noninvasively monitoring these pathological alterations in the liver. Liver biopsy for histological analysis still remains the gold standard for evaluation of liver disease. However, because the use of liver biopsy has several limitations, including sampling error,3 complications,4, 5 discomfort, and ethical limitations of performing repeated liver biopsies, this practice for monitoring therapeutic response has been markedly reduced. Furthermore, because biopsy is an assay that provides only local anatomical information without providing functional assessment of the whole organ, it is frequently combined with a functional assay such as serum levels of different markers. Serum marker concentrations, however, are not specific to changes in the liver and cannot accurately discriminate between different disease stages.6, 7
In the last decade, advances in radiological imaging such as computed tomography (CT) and magnetic resonance imaging (MRI) have provided the clinician with a number of diagnostic modalities to noninvasively assess liver diseases and regeneration.8–11 However, existing diagnostic imaging techniques provide only limited evaluation of tissue characteristics beyond morphology. Because liver pathologies affect blood flow, perfusion imaging of the liver has been suggested to improve the sensitivity and specificity of liver diagnostic tools.12 The utility of hepatic perfusion characterization relies on the resolution of each component of its dual blood supply, the portal vein and the hepatic artery, because contributions from each are altered in many liver diseases.12 Changes in relative arterial and portal venous blood flow are known to be associated with progression of cirrhosis.13, 14 In primary and metastatic liver malignancies, there is a relative increase in arterial blood supply to the tumor.15 In addition, physiological changes such as regeneration are also associated with hemodynamic changes.16–18
Blood perfusion measurements using MRI can potentially overcome limitations posed by other imaging techniques, such as poor spatial resolution in radionuclide studies, lack of reproducibility in Doppler ultrasound and radiation exposure using CT. Today, in order to acquire perfusion images in both CT and MRI, intravenous administration of a contrast agent is necessary. Good separation of the arterial from the portal phase requires a high temporal resolution which comes at the cost of a reduced spatial resolution. Recently, we demonstrated the feasibility of functional MRI (fMRI) combined with hypercapnia and hyperoxia for monitoring changes in liver perfusion and hemodynamics without contrast agent administration.19 The fMRI method is based on changes in proton signals from tissues that are adjacent to blood vessels containing paramagnetic deoxyhemoglobin.20 Thus, changes in deoxyhemoglobin levels cause local magnetic field susceptibility gradients, which can be detected by T2*-weighted gradient echo MRI as signal intensity (SI) changes. These changes are related to changes in oxygen saturation, blood flow, and blood volume.21 Using this technique, we established that in the liver, the changes in the magnitude of both ΔSco2 and ΔSo2 (percent change in SI due to hypercapnia and hyperoxia, respectively) reflect changes in total blood volume, whereas changes in sign from negative to positive ΔSco2 values reflect relative changes in portal/arterial flow ratio.19
In this study, in an effort to expand the use of this technique to monitor liver pathology, we evaluated changes in fMRI SI during fibrosis, partial liver resection, and hemorrhage in in vivo rat and mice models. Results from these studies could provide the basis for the development of fMRI combined with hypercapnia and hyperoxia as a noninvasive, rapidly responding monitoring method for evaluation and follow-up of patients with liver pathologies and for assessing therapeutic responses to different interventions.
ΔSco2, signal intensity change due to hypercapnia; ΔSo2, signal intensity change due to hyperoxia; fMRI, functional MRI; MRI, magnetic resonance imaging; PHx, partial hepatectomy; SI, signal intensity; TAA, thioacetamide.
Materials and Methods
All experiments were performed in accordance with the guidelines of the Animal Care and Use Committee of the Hebrew University (National Institutes of Health approval number OPRR-A01-5011). Adult male Sprague-Dawley rats (250–300 g) were used for hypovolemia, partial hepatectomy, and acute fibrosis model experiments. For the moderate fibrosis model, 2-month-old and 4-month-old multidrug resistance protein 2 gene knockout (Mdr2−/−) mice and equivalent control (FVB/NJ) mice22 were analyzed. Mdr2−/− mice are deficient in the canalicular phospholipid flippase and spontaneously develop liver injury, due to the absence of phospholipid from bile, which leads to the formation of periportal biliary fibrosis at an early age.23
MRI Analysis Technique.
MRI experiments were performed on anesthetized rodents (pentobarbital, intraperitoneally at 30 mg/kg body weight) in a horizontal 4.7-T Biospec spectrometer (Bruker Medical, Ettlingen, Germany) using a 7-cm birdcage coil for rats and a 3.5-cm birdcage coil for mice. Coronal and axial T1-weighted spin echo images of the whole liver were acquired for alignment and to determine liver volume (repetition time = 500 ms; echo time = 18 ms).
Hepatic perfusion and hemodynamics were evaluated from T2*-weighted gradient echo images (repetition time = 147 ms; echo time = 10 ms; flip angle = 30; field of view = 5.8 cm [rats] or 3 cm [mice]; 256 × 128 pixels; in-plane resolution = 230 μm [rats] or 117 μm [mice]; five slices with slice thickness = 1.5 mm [rats] or 1 mm [mice]; spectral width of 25,000 Hz; four averages; 65 seconds/image) acquired during breathing of air, air-CO2 (95% air and 5% CO2), and oxygen-CO2 (95% oxygen and 5% CO2) as described.19, 24 Five repeats were acquired at each gas mixture. Zero filling of k-space data was applied, resulting in a matrix of 256 × 256 pixels.
For all experiments, we performed baseline MRI liver acquisition scanning with the different gas mixtures, as mentioned above.
Effect of hypovolemia:
A catheter was inserted into the femoral artery for measurements of blood pressure and for bleeding the rats (controlled hemorrhage). Blood samples were taken to assess pH, hemoglobin, and hematocrit in order to confirm hemorrhagic status. After baseline, we withdrew approximately 15% (4.8 ± 0.3 mL) (n = 5) or 35% (10.5 ± 2.5 mL) (n = 7) of the total blood volume (≈0.5 mL/minute), and then another MRI scan was acquired. Rats were sacrificed 6 hours after bleeding, and livers were fixed for histological evaluation.
Effect of partial hepatectomy:
Resection of 70% of the total liver mass was performed in rats according to Higgins and Anderson25 under light anesthesia (diethyl ether) by removing the median and left lateral lobes (n = 12). Rats were imaged every other day, on days 0-10 after partial hepatectomy (PHx) and additional livers were fixed for histological analysis.
Effect of fibrosis:
Acute liver fibrosis was induced in rats by intraperitoneal administration of thioacetamide (TAA) (Sigma, Rechovot, Israel) at 0.2 mg/g body weight twice weekly for 8 weeks.26 Under anesthesia, rats were scanned weekly during the 8 weeks of TAA administration (n = 7 for each time point). For the detectability assessment of mild fibrosis, 2-month-old and 4-month-old Mdr2−/− mice (n = 8) and equivalent FVB/NJ mice (n = 5) were scanned with the different gas mixtures. The livers were obtained for histological evaluation.
Animals were sacrificed at parallel time points to MRI scans and their livers were fixed in formalin. Sections (4 μm) were cut and stained with hematoxylin and eosin (H&E). For fibrosis assessment, sections were also stained with Masson's Trichrome (collagenous tissue). Immunohistochemistry was performed on paraffin-embedded sections stained for myeloperoxidase using a 1:50 dilution of monoclonal antibody-1 (NeoMarkers, Fremont, CA).
In brief, the liver boundary visualized in each slice was outlined using image-processing software (NIH ImageJ 1.37r; National Institutes of Health, Bethesda, MD). To convert the number of liver pixels to an area, we multiplied by the factor [(field of view)2/(matrix)2]. The total liver volume was calculated as the summed area of all slices, multiplied by the slice thickness. For each rat, the liver volume was expressed as a percentage of the preoperative liver volume.
The MRI hemodynamics data were analyzed on a PC computer with home-written software using Interactive Data Language programming (ITT Visual Information Solutions, Boulder, CO). Maps of the mean SI values for each pixel obtained during inhalation of the different gases (Sair, Sco2, and So2) were calculated from the average of four repeats for each gas (discarding the repeats obtained during gas changes19). The percentage change in the intensity of the MRI signal induced by hypercapnia (ΔSco2) and by hyperoxia (ΔSo2) was calculated according to the following equations:
Data are presented in color maps. Results are expressed as means ± standard deviation (SD). Regions of interest were defined using anatomical images and included the whole liver or back muscle excluding big blood vessels and flow artifacts.19 Mean values were calculated from these regions of interest from the number of animals (n) as indicated, using three slices per animal.
Differences between groups were identified by the unpaired Student t test. A P value of less than 0.05 was considered statistically significant.
Effect of Hypovolemia on ΔSco2 and ΔSo2.
We evaluated the ability of fMRI combined with hyperoxia and hypercapnia to assess the liver response to acute hemodynamic changes. It has previously been suggested that macrocirculatory and microcirculatory disturbances of liver perfusion during and after acute hemorrhage are major contributors to hepatic failure, which consequently is associated with increased mortality.27, 28 Mean ΔSo2 values (the change induced by hyperoxia) obtained subsequent to 15% and 35% hemorrhage of total blood volume was reduced by 40% and 82%, respectively, compared to normovolemia (Fig. 1A; P < 0.001). Mean values of ΔSco2 (the change induced by hypercapnia) were significantly less negative by 58% and 84%, respectively, compared to normovolemia (Fig. 1B; P < 0.001). The change in ΔSo2 and ΔSco2 values after hemorrhage correlated with the percentage of blood loss (Fig. 1; R = 1). These findings suggest that the decrease in blood volume diminished the change brought by gas saturation. The pathological changes in histological liver sections could be detected only 6 hours after hemorrhage and not in sections from animals sacrificed earlier (data not shown). After hemorrhagic shock in rats (withdrawal of 35% of total blood volume), liver damage was manifested by fatty changes, inflammatory cell infiltration, and diffuse acidophilic bodies (Fig. 1C). Additionally, examination of these liver sections detected numerous apoptotic nuclei.
Effect of PHx and Regeneration on ΔSco2 and ΔSo2.
We used MRI to follow liver regeneration kinetics subsequent to 70% PHx. Average liver volume reached 115% of its initial size on day 4 (Fig. 2). In addition to the anatomical changes occurring during liver regeneration, liver hemodynamics was greatly affected until regeneration concluded. Two days after 70% PHx, mean ΔSo2 values declined significantly (prehepatectomy value, 29.5 ± 6.2%; posthepatectomy value, 0.4 ± 6%, P < 0.01), illustrating a decrease in liver perfusion and blood content. Once regeneration progressed, from day 4 onward, ΔSo2 values gradually returned to normal (Fig. 3A,C). In addition, immediately after PHx, mean ΔSco2 values increased and turned positive (prehepatectomy value, −21.4 ± 6%; posthepatectomy value, 11.2 ± 10%, P < 0.05) signifying a change in the source of liver nourishment (Fig. 3B,C). Because there is a growing need for oxygenated blood in the liver, the hepatic artery becomes the main source perfusing the liver with oxygenated blood,17 thus generating positive ΔSco2 values. Mean ΔSco2 values were likewise restored with the conclusion of regeneration (Fig. 3B). Histological sections stained with H&E at parallel time points to the MRI scans confirmed the decrease in vascular density on posthepatectomy days 2 to 4 (Fig. 3E-H).
Effect of Fibrosis on ΔSo2 and ΔSco2.
Liver fibrosis increases intrahepatic resistance to blood flow through scar formation, reduces perfusion, and induces shunting.13, 16 Anatomical images (T1-weighted) of fibrotic rat livers had higher SI than normal livers due to extracellular deposition (Fig. 4A, top; Fig. 5E-F). In addition, liver volume of TAA-treated rats was significantly reduced (Fig. 4A). Two weeks after TAA administration, ΔSo2 values were much higher and ΔSco2 values became much lower (109 ± 40%, −44.6 ± 10% respectively, P < 0.01) compared to ΔS values from normal rat livers, suggesting augmented flow to the liver as a result of the inflammatory process (Fig. 4). Inflammation was confirmed by histological sections revealing parenchymal neutrophil infiltration identified by immunostaining with anti-myeloperoxidase antibodies and by H&E staining (Fig. 5B,H). After the first 2 weeks of TAA treatment, with the development of massive fibrosis, ΔSo2 values decreased (1 ± 3%, P < 0.01) and ΔSco2 values gradually turned positive (1.9 ± 6%, P < 0.01) (Fig. 4B). Because the development of liver fibrosis in the TAA rat model was rapid and did not resemble the kinetics of human disease, an additional mouse model was analyzed. It has been previously shown that Mdr2−/− mice develop liver fibrosis already at the age of 2 weeks.23 The level of fibrosis in the 2-month-old Mdr2−/− mice was higher than in the 4-month-old Mdr2−/− mice as assessed by a professional pathologist (16.3% and 11.7% respectively; Fig. 6K,L). Indeed, both ΔSo2 and ΔSco2 were significantly attenuated in both 2-month-old and 4-month-old Mdr2−/− mice compared to controls (Fig. 6; P < 0.01). In addition, as opposed to the homogenous ΔS normal liver maps, the Mdr2−/− liver ΔS maps demonstrated an inhomogenous pattern with higher reactivity in the big vessels. Histological staining confirmed the grade of liver fibrosis in this model (Fig. 6K,L). When we analyzed the time-course of the SI change in these mice, we observed a delay in the liver response to CO2 in the 4-month-old Mdr2−/− mice and a bigger delay for both CO2 and O2 in the 2-month-old Mdr2−/− mice (Fig. 6N). These findings suggest that during liver fibrosis, there is a decrease in the dynamic responsiveness of the liver to hyperoxic and hypercapnic challenges compared to healthy livers. These changes may reflect structural and functional alterations of the liver sinusoids and vasculature, which affects hepatic vascular reactivity.
There is a growing need for noninvasive tools to diagnose and monitor pathological changes in the liver. These pathological processes are frequently accompanied by changes in hepatic perfusion and hemodynamics.13, 29 The fMRI method uses the intrinsic contrast of deoxyhemoglobin, and therefore allows one to follow blood perfusion without contrast agent administration. Only a small number of previous studies have used a similar method of O2 enrichment to determine liver oxygenation level.17, 30–32 By assessing the response to CO2 inhalation, we acquired additional information about the characteristics of the hepatic blood flow. Recently, we reported the applicability of fMRI during hypercapnia and hyperoxia as a tool for monitoring changes in liver perfusion and hemodynamics.19 This study complements our previous findings by examining the feasibility of this method to evaluate hemodynamic changes that occur during liver pathologies. This fMRI method improves liver perfusion diagnostic and monitoring abilities for different pathologies, for example, early perfusion changes occurring in the hemorrhagic liver, additional evidence of the arterial flow following PHx, and illustration of the inflammatory process at the early stages of fibrosis and the hemodynamic unresponsiveness during the late fibrotic progression. This method was sensitive enough to detect even mild fibrosis.
In a previous study, we demonstrated that during CO2 enrichment there is an increase in portal flow compared to arterial hepatic flow.19 The resultant higher deoxyhemoglobin levels produced a decrease in fMRI SI, which is illustrated by negative ΔSco2 values. Thus, liver ΔSco2 values are sensitive to portal versus arterial perfusion. The signal change induced by hyperoxia (ΔSo2) signifies vascular density and tissue perfusion.19, 33 Normally, the liver receives 25% of the cardiac output and contains 15% of the total blood volume of the body.34 This fact and the hypervascularity of the liver explain the high ΔSo2 values in the liver compared to the muscle, as observed in our previous study.19
When we reduced blood volume by withdrawing blood, the mean ΔSo2 values declined to a level similar to that observed in the muscle. Nevertheless, during hypovolemia, mean ΔSco2 values remained negative, suggesting that portal flow relative to systemic flow, and hence deoxy/oxyhemoglobin ratio in the blood nourishing the liver, did not change considerably. In this study, we found a very good correlation between the level of blood loss and the changes in ΔSo2 and ΔSco2 values. The change in ΔSo2 and ΔSco2 values induced by hemorrhage signifies the decrease in blood volume and vascular reactivity as a result of the physiological response of the body to hypovolemia (for example, redistribution of blood to vital organs and reduced responsiveness of blood vessels, which are already dilated or vasoconstricted to their maximum level). Thus, the novelty of this technology enabled us to monitor very early responses of liver hemodynamics to hemorrhagic shock which were not detectable at early time points by analysis of histology and liver enzymes. We further extended this work by showing the ability of the fMRI method to assess liver outcome after hemorrhagic shock and subsequent to different resuscitation approaches targeted to improve liver hemodynamics and outcome. We showed that adrenaline may be preferable to Ringer lactate solution as an initial measure to attenuate liver injury following hemorrhagic shock.35
The advantages of using a noninvasive imaging technique, such as MRI, to monitor liver regeneration in rats has previously been demonstrated.9 In the process of liver regeneration after PHx, liver hemodynamics is immensely affected until regeneration concludes. In this study, increased mean ΔSco2 values were observed after PHx. Previous studies reported hemodynamic alterations during PHx.17, 18 Because more oxygenated blood is required due to an increased O2 consumption during liver regeneration, arterial blood flow increases significantly as measured by flow meters.17 This would explain the positive mean values of ΔSco2 obtained during the initial period of liver regeneration (days 1-5), as the deoxy/oxyhemoglobin ratio changes in the tissue. Because the proliferation of hepatocytes precedes endothelial cell proliferation,16, 36 there is a relative reduction in the perfusion of the liver parenchyma, which is evident in the reduced mean ΔSo2 values during the initial period of liver regeneration (Fig. 3). Within 2 weeks, values of ΔSco2 and ΔSo2 normalized, indicating that the acute regenerative phase had completed. Therefore, we were able to demonstrate and follow the dynamic changes in the whole liver that occur during the regenerative process. This technique can further be used in the future to monitor, both experimentally and clinically, regenerative liver processes and to gain information regarding the functional properties of the liver. For example, this would enable monitoring of different transgenic and knockout mice in order to investigate the importance of specific genes during liver regeneration.
Hepatic fibrosis is associated with total derangement of the normal architecture due to the presence of regenerative nodules and fibrous bands or septae running between them. Thus, fibrosis results in reduced functionality and obliteration of hepatic vasculature. The liver hemodynamic status, therefore, includes changes involving arterializations of the liver, formation of intrahepatic shunts (between the branches of the hepatic artery, the portal vein, and the hepatic veins), and development of a hyperdynamic circulatory state.13, 14 MRI techniques, such as dynamic gadolinium-enhanced MRI, enable only evaluation of progressive stages of liver disease.37, 38 Using the fMRI technique, we were able to identify hemodynamical changes that occur early during the acute liver fibrotic process. Within 1 week after TAA administration, we were able to detect alterations in ΔSco2 and ΔSo2 responsiveness resulting from inflammation. With progression of the fibrotic process, ΔSco2 values gradually turned positive and ΔSo2 values decreased, indicating a decrease in the dynamic response of the liver to hyperoxic and hypercapnic challenges compared to healthy livers. Our method proved to be sensitive enough to detect changes occurring as a result of mild fibrosis by using the Mdr2−/− mouse as a model. The change illustrated in the livers of these mice occur both by attenuation of ΔS responsiveness and homogeneity of the liver parenchyma. These changes reflect the structural and functional alterations in the liver vasculature, as described.13, 14 Doppler ultrasound studies have been conducted in order to assess portal hypertension, hepatic flow, and response to treatment; however, this technique is not as sensitive and accurate as MRI and CT.38, 39 Subsequently, Doppler ultrasound is limited to offering only evaluations of blood flow in specific blood vessels, whereas the fMRI method we present here provides a comprehensive and integrative view of liver hemodynamics.
The fMRI method is also applicable and informative when performed on mice.40 Previously, we were able to assess the inflammatory process induced in the liver of caspase-8 knockout mice resulting from PHx.40 There should be no concerns regarding the changes in SI that could be observed in humans because blood volume is higher than in rats. Thus, changes in signal intensity are more intense and could also be observed with 1.5-T scanners that are currently acceptable for human use, unlike the 4.7-T scanners used in this study. Liver fMRI in humans can be limited by a breathing-related motion artifact which is very critical for achieving significant ΔS maps. The Echo-planar MR imaging technique is used for brain fMRI in humans; however, it still suffers from lower imaging resolution. We are now in the process of implementing this method in human subjects and we have succeeded in producing significant ΔS maps using Echo-planar imaging (unpublished data).
In summary, fMRI combined with hyperoxia and hypercapnia offers a unique opportunity to noninvasively monitor the functionality and structural changes in the liver. Our data in rats and mice provide comprehensive evidence for the use of this method to follow a variety of physiological and pathological processes such as liver fibrosis and regeneration and the liver response to resection or acute bleeding. There are a number of diagnostic applications by means of MRI for anatomical and functional liver evaluation that are considered supreme to CT and ultrasound.39, 41–45 The offered fMRI method provides a more comprehensive view of the hemodynamic status of the liver compared to other existent diagnostic tools that are either limited to specific blood vessels or are invasive. In addition, the fMRI method is sensitive to very early hemodynamic changes that could not be demonstrated otherwise. Combining the fMRI method with the standard MRI diagnostic protocol will hopefully enable a thorough and accurate diagnosis with additional information regarding functionality properties. Further studies are required to investigate the capability of this technique to evaluate and follow up fibrosis, liver tumors, and response to different therapeutic interventions in humans.
We thank Dr. Eli Pikarsky and his laboratory, Dr. Rinnat M. Porat, Dr. Ilan Stein, and Shafika Kasem (Hadassah University Hospital) for the helpful assistance with the histological sections.