Hepatic bile acid transport increases in the postprandial state: A functional 11C-CSar PET/CT study in healthy humans

Background & Aims It is not known how hepatic bile acids transport kinetics changes postprandially in the intact liver. We used positron emission tomography (PET)/computed tomography (CT) with the tracer [N-methyl-11C]cholylsarcosine (11C-CSar), a synthetic sarcosine conjugate of cholic acid, to quantify fasting and postprandial hepatic bile acid transport kinetics in healthy human participants. Methods Six healthy human participants underwent dynamic liver 11C-CSar PET/CT (60 min) during fasting and from 15 min after ingestion of a standard liquid meal. Hepatobiliary secretion kinetics of 11C-CSar was calculated from PET data, blood samples (arterial and hepatic venous) and hepatic blood flow measured using indocyanine green infusion. Results In the postprandial state, hepatic blood perfusion increased on average by 30% (p <0.01), and the flow-independent hepatic intrinsic clearance of 11C-CSar from blood into bile increased by 17% from 1.82 (range, 1.59–2.05) to 2.13 (range, 1.75–2.50) ml blood/min/ml liver tissue (p = 0.042). The increased intrinsic clearance of 11C-CSar was not caused by changes in the basolateral clearance efficacy of 11C-CSar but rather by an upregulated apical transport, as shown by an increase in the rate constant for apical secretion of 11C-CSar from hepatocyte to bile from 0.40 (0.25–0.54) min−1 to 0.67 (0.36–0.98) min−1 (p = 0.03). This resulted in a 33% increase in the intrahepatic bile flow (p = 0.03). Conclusions The rate constant for the transport of bile acids from hepatocytes into biliary canaliculi and the bile flow increased significantly in the postprandial state. This reduced the mean 11C-CSar residence time in the hepatocytes. Lay summary Bile acids are important for digestion of dietary lipids including vitamins. We examined how the secretion of bile acids by the liver into the intestines changes after a standard liquid meal. The transport of bile acids from liver cells into bile and bile flow was increased after the meal.


Introduction
Bile acids are produced in the hepatocytes and secreted across the apical membrane into the biliary canaliculi by active transport 1 and flow with bile into the gut. Here, bile acids are important for absorption of dietary lipids and lipophile substances. 2 The intestinal uptake of bile acids, which primarily takes place in the terminal ileum, is highly efficient and facilitated by active transmembrane transport proteins. 3 After absorption, the bile acids are returned by the portal venous system to the liver where they are removed from the blood and subsequently secreted into the biliary tree. This enterohepatic circulation of bile acids depends strongly on an intact and adaptive transhepatic bile acid transport and secures a high bile acid concentration in the gut with little de novo synthesis. 4,5 In view of the physiological importance of bile acids, surprisingly little is known about how food intake affects the hepatobiliary bile acid transport kinetics. Soloway and Schoenfield 6 collected bile from patients through a T-tube catheter placed in the common bile duct after cholecystectomy and reported a 40% increase in hepatic bile acid secretion and a 46% increase in bile flow after a morning meal. Based on intravenously administered radiolabelled bile acids, they further concluded that the increase was caused by increased transhepatic transport of bile acids from blood to bile rather than by hepatic de novo synthesis. However, recent surgery to the biliary system with cholecystectomy together with the collection of bile from an invasive catheter that may disturb the enterohepatic circulation of bile acids could have altered the physiology of bile formation and flow.
Positron emission tomography (PET) with the radiolabelled synthetic sarcosine conjugate of cholic acid [N-methyl- 11 C] cholylsarcosine ( 11 C-CSar) enables in vivo quantification of the separate steps of the hepatobiliary transport of bile acids from blood to bile by external detection and thus without any confounding effects of invasive procedures. [7][8][9][10] In the present study, we used PET to quantify the hepatobiliary secretion kinetics of examine whether these changes could be related to altered kinetics of the sinusoidal (basolateral) uptake and/or biliary (apical) secretion of the tracer.

Human participants
Six healthy participants (2 women and 4 men, mean age 27 years) with no history of liver disease or daily intake of drugs were included through internet advertisement (Table 1). Routine blood tests including alanine aminotransferase, alkaline phosphatase, bilirubin, gamma-glutamyltransferase, sodium, potassium, albumin, creatinine, C-reactive protein, leucocytes, haemoglobin, thrombocytes, international normalised ratio, glucose, and activated partial thromboplastin time were measured on the day of the experiment and were normal in all participants.

Ethics
The study was approved by the Central Denmark Region Committees on Health Research Ethics (1-10-72-272-16) and conducted in accordance with the Helsinki II Declaration. Written informed consent was obtained from all participants. Pregnancy was ruled out by a negative human chorionic gonadotropin urine test in all female participants. The participants were compensated, receiving 1,000 Danish kroner (approximately 130 euro) for participation. No complications to the procedures were observed. The maximum radiation dose received by the participants was 4.3 mSv.

Study design
The experiments were conducted in the morning at the PET Centre after an overnight fast (minimum 8 h). First, catheters (Venflon, Becton Dickinson, Kongens Lyngby, Denmark) were placed in both antebrachial veins for administration of the 11 C-CSar tracer and indocyanine green (ICG), as well as in a hepatic vein via the right femoral vein (Torcon Advantage, Cook Inc., Bjaeverskov, Denmark) and in a radial artery for blood sampling (Artflon, Becton Dickinson, Kongens Lyngby, Denmark).
The first 60-min 11 C-CSar PET scan was performed with the participant fasting. To avoid signal interference from the first to the second scan, the participant rested for 1 h after the end of the first scan. After this rest, the participant ingested a 500-ml standard liquid meal based on egg and milk produced by the central kitchen at Aarhus University Hospital on the day of the experiment. The energy content was 4,291 kJ, with 1,417 kJ from protein (33%), 1,394 kJ from fat (32%), and 1,468 kJ from carbohydrates (35%). Previous studies have shown that the hepatic blood flow increases shortly after ingestion of a meal before slowly reverting back to normal over the following hours. [11][12][13] Therefore, the second 60-min 11 C-CSar PET scan was initiated 15 min after ingestion of the standard meal. 11 C-CSar PET/CT recording The PET/computed tomography (CT) studies were performed using a Siemens Biograph 64 Truepoint PET/CT camera (Siemens AG, Ballerup, Denmark). Participants were studied in the supine position, and a tomogram was used to place the individual so that the liver and the bile ducts were in the centre of the 21-cm field of view of the PET camera. A mean dose of 11 C-CSar 129 MBq (range, 59-205 MBq) was injected intravenously during the initial 20-30 s of a 60-min PET recording. 11 C-CSar was produced on-site at the Department of Nuclear Medicine & PET Centre as previously described. 14 PET data were recorded in list mode and reconstructed using attenuation-weighted ordered-subset expectation maximisation with resolution recovery (TrueX) with 4 iterations, 21 subsets, a 336 × 336 × 109 matrix, and a 2-mm Gauss filter. The final PET image voxel size was 2 × 2 × 2 mm 3 , and the time frame structure was 9 × 10 s, 10 × 45 s, and 17 × 3 min. PET measurements were corrected for radioactive decay back to the start of the PET recording.
PET images were analysed using the PMOD software (PMOD Technologies Ltd, Zürich, Switzerland). Regions of interest were drawn in the liver tissue in adjacent planes and combined to produce a volume of interest that was used to generate the time course of the liver tissue concentration of 11 C-CSar. The total liver volume (V liver ) was estimated using the iso-contour tool from the time period after arterial peak had passed (approximately 70 s) and before 11 C-CSar appeared in the extrahepatic bile ducts (approximately 180 s). In this period, the hepatic concentration of 11 C-CSar was greater than the arterial concentration, and the threshold for including the hepatic regions was set to 10% above the arterial peak value in this period.
Hepatic blood flow Hepatic blood flow (ml blood/min) was measured during both PET scans using a constant intravenous infusion of ICG, measurements of plasma concentrations of ICG in arterial and hepatic venous blood, and Fick's principle and corrected for nonsteady state. 15,16 The hepatic blood perfusion, Q (ml blood/min/ ml liver tissue), was calculated as the hepatic blood flow divided by V liver .

Hepatobiliary 11 C-CSar kinetics
During both PET recordings, blood samples were taken from the radial artery and hepatic vein (more frequently in the beginning than later during the scan) for measurements of the blood concentration of 11 C-CSar. 14 The concentration of 11 C-CSar in portal venous blood was calculated from the arterial blood concentration using a model for the transfer of tracer through the splanchnic bed. 9,17 This model assumes no loss or gain of tracer in the prehepatic splanchnic bed and includes a tracer specific time constant, b, which was estimated for 11 C-CSar in fasting pigs to a mean value of 13 s (95% CI, 6.0-18.0 s). 8 The flow-weighted Research article mixed input of 11 C-CSar to the liver, C in (t), was calculated from the arterial and the modelled portal venous concentrations of 11 C-CSar using a hepatic arterial flow fraction of 0.25. 18,19 The kinetic analysis was based on the model illustrated in Fig. 1 and provided quantitative description of the different steps of the hepatic transport of 11 C-Sar from blood to bile, as previously described. 9,20 The hepatic extraction fraction, E(t), was calculated using C in (t) and the output concentration of 11 C-Csar in the hepatic vein, C out (t). C out (t) was corrected for non-steady state using individual estimates of mean hepatic blood transit time, T, calculated as the difference between the peak time of the 11 C-CSar concentration in hepatic input and output.
The initial extraction fraction, E 0 , was calculated using the recordings of the first minute (first pass) after the bolus injection where potential backflux of 11 C-CSar from hepatocytes to blood can be ignored.
The AUCs of C in (t) and C out (t) for the first 50 min after administration of 11 C-CSar were used to calculate E AUC , which was used as a measure of the steady-state hepatic extraction fraction. 9 The permeability surface area product of the hepatocyte plasma membrane, PS mem , which is a measure of the efficacy of the transport of bile acids across the sinusoidal membrane from blood to hepatocytes, was calculated as 21 The flow-independent intrinsic clearance, Cl int , which is a measure of the overall hepatic transport efficacy from blood to bile, was calculated as 22 The kinetic model shown in Fig. 1 consists of well-mixed compartments of blood, hepatocytes, and intrahepatic bile ducts. The unidirectional clearance of 11 CSar from blood into hepatocyte, K 1 , and the rate constants k 2 , k 3 , and k 5 for the transport between the compartments were estimated by nonlinear regression using C in (t) as input function and hepatic tissue concentration as output function. Using the previously published normal values for the rate constants as a starting point in the regression, the fitting was consistent and reproducible without the use of further assumptions. 9,20 This enabled estimation of rate constants for the separate transport steps and calculation of mean hepatocyte residence time of 11 C-Csar 9 : T hep is the average time 11 C-CSar molecules reside in the hepatocytes before being either secreted into bile (k 3 ) or transported back to blood (k 2 ). For k 3 >> k 2 , T hep characterises the time a 11 C-CSar molecule uses to traverse the hepatocyte from blood to bile. We also calculated bile flow (F bile = k 5 V liver V bile , where V liver is the total liver volume as defined above and V bile is the fraction of intrahepatic bile ducts in the liver tissue) and concentration ratios between blood, hepatocyte, and bile, as previously described. 9 Endogenous bile acids and fibroblast growth factor 19 Before and after each of the 2 PET scans,

Statistical analysis
All data were normally distributed and are therefore presented as mean (95% CI). Comparisons between baseline and postprandial state were examined using paired t tests, and a p value <0.05 was considered to indicate a statistically significant difference between the fasting and postprandial states. Statistical analyses were performed using STATA software, release 13 (StataCorp, College Station, TX, USA). The number of participants was based on previous experience from functional PET studies. Fig. 2 shows examples of the time courses of the concentrations of 11 C-CSar in the fasting and postprandial states in the same participant standardised for injected dose of 11 C-CSar and body weight. The concentration of 11 C-CSar in the liver tissue increased rapidly after 11 C-CSar administration and peaked at comparable levels in the fasting and postprandial states, whereas the subsequent decrease, representing hepatobiliary secretion and bile flow out of the liver, was more rapid after food intake than during fasting. In the postprandial state, hepatic venous blood had a higher concentration of 11 C-CSar, likely as a result of increased hepatic blood flow, but not enough to affect the hepatic extraction fractions as described in the next section.

Results
Kinetics for the uptake of 11 C-CSar from blood to hepatocyte Neither E 0 nor E AUC was statistically significantly different when comparing the postprandial and fasting states (  The model describes compartments of blood, liver tissue, and intrahepatic bile within the liver-VOI. The exchange of 11 C-CSar between the compartments is described by rate constants: K 1 , unidirectional clearance of 11 C-CSar from blood to hepatocytes (ml blood/min/ml liver tissue); k 2 , rate constant for the backflux from hepatocytes to blood (min −1 ); k 3 , rate constant for secretion from hepatocytes to intrahepatic bile (min −1 ); and k 5, rate constant for flow of 11 C-CSar to extrahepatic bile (min −1 ). 11 C-CSar, [N-methyl-11 C]cholylsarcosine; liver-VOI, liver volume of interest.
(2.36-5.80) ml blood/min/ml liver tissue in the postprandial state (p = 0.51). PS mem was more than 3 times higher than hepatic blood perfusion, typical for high extraction compounds and underlining the efficient transport of 11 C-CSar from blood to hepatocyte.
As shown in Table 2, mean hepatic blood perfusion increased by 30% in the postprandial state (p <0.01). With a high PS mem , hepatic clearance of 11 C-CSar from blood will increase with flow, 22 and in accordance, K 1 , the flow-dependent clearance of tracer from blood to hepatocyte, increased proportionally from 0.95 (0.66-1.23) ml blood/min/ml liver tissue to 1.24 (0.87-1.61) ml blood/min/ml liver tissue in the postprandial state (p = 0.003).
The meal did not change PS mem , but it did increase the flowindependent hepatic intrinsic clearance of 11 C-CSar from blood to bile (Cl int ) by on average by 17% from a mean baseline value of 1.82 (1.59-2.05) ml blood/min/ml liver tissue to a mean postprandial value of 2.13 (1.75-2.50) ml blood/min/ml liver tissue (p = 0.042). This illustrates that the overall transport capacity of 11 C-CSar from blood to bile increased postprandially, an increase that was independent of the increased hepatic blood flow and explained by an increased apical transport of 11 C-CSar from hepatocytes to bile canaliculi (see below).
Kinetics for the transport of bile acids from hepatocyte to bile As seen in Table 2, the rate constant for backflux of 11 C-CSar from hepatocyte to blood, k 2 , did not change postprandially (p = 0.55), whereas the rate constant for apical secretion from hepatocytes to bile, k 3 , increased 68% from mean 0.40 (0.25-0.54) min −1 to mean 0.67 (0.36-0.98) min −1 (p = 0.03) (Fig. 3). This is in accordance with the observed increase in Cl int .
The mean hepatic residence time of 11 C-CSar in the hepatocytes was 2.79 (1.57-4.00) min during fasting and decreased significantly to 1.90 (0.57-3.24) min in the postprandial state (p = 0.03) because of increased k 3 . As shown in Table 2, this secured unchanged concentration ratios between blood and hepatocyte  Values in bold denote statistical significance. Q, hepatic blood perfusion (ml blood/min/ml liver tissue); E 0 , unidirectional hepatic extraction fraction of 11 C-CSar; E AUC , hepatic extraction fraction of 11 C-CSar using AUC of input and output at 50 min; PS mem , permeability surface area product of 11 C-CSar (ml blood/min/ml liver tissue); Cl int , hepatic intrinsic clearance of 11 C-CSar (ml blood/min/ml liver tissue); K 1 , unidirectional clearance of 11 C-CSar from blood to hepatocytes (ml blood/min/ml liver tissue); k 2 , rate constant for backflux of 11 C-CSar from hepatocytes to blood (min −1 ); k 3 , rate constant for the secretion of 11 C-CSar from hepatocytes to bile (min −1 ); k 5 , rate constant for flow of 11 C-CSar in bile out of the liver-VOI (min −1 ); T hep (min), which is the average time 11 C-CSar molecules reside in the hepatocytes; C hep /C in , the ratio between the steady-state concentration of 11 C-CSar in hepatocytes and blood; C bile /C hep , the ratio between intrahepatic bile and hepatocytes; 11 C-CSar, [N-methyl-11 C]cholylsarcosine; liver-VOI, liver volume of interest.
Research article (p = 0.45) and between hepatocyte and intrahepatic bile (p = 0.41). Fig. 5 shows a large interindividual variation in the postprandial plasma concentrations of bile acids and FGF-19. As no differences were found between the 2 concentrations measurements of FGF-19 during fasting (at the start and the end of the fasting experiment, p >0.10 for all), the mean of the 2 fasting values (venous and arterial separately) were used as a baseline for comparison with postprandial values. In the postprandial state, the concentration of endogenous bile acids increased significant after 20 min, whereas only the 60-min value was significantly different for FGF-19 (Fig. 5).

Discussion
This is the first in vivo quantification of the postprandial changes in the hepatobiliary transport of bile acids from blood to bile. The main observation was an increase in the apical transport rate constant (k 3 ) by two-thirds in the postprandial state as compared with the fasting state. This occurred during the period 15 min to 1.25 h after food intake and significantly reduced the mean hepatic residence time of bile acids. The hepatobiliary kinetics during fasting (Table 2) were in the same order of magnitude as our previously published data, and  with low interindividual variability, indicating the stability of the methodology used and described in detail elsewhere. 9,20 Fasting PS mem was high and did not change after the meal, and in accordance with this, E o was close to 1.0 both before and after the meal. Interestingly, Cl int increased in the postprandial state (from 1.82 to 2.13 ml blood/min/ml liver tissue) which suggests an increase in the transport capacity of bile acids from hepatocytes into biliary capillaries briefly after a meal as quantified by the rate constant for secretion, k 3 , which increased from a mean fasting value of 0.40 min −1 to a mean postprandial value of 0.67 min −1 . This means that in the postprandial state, 67% of hepatocellular 11 C-CSar was secreted into biliary capillaries every minute in contrast to only 40% during fasting. The increase in k 3 reduced the hepatic residence time so that the average 11 C-CSar molecule was 1.9 min (0.57-3.24 min) to traverse the hepatocyte in the postprandial state, in contrast to 2.79 min (1.57-4.00 min) in the fasting state. In accordance with bile flow being largely regulated by bile acids, 23 the rate constant for bile flow, k 5 , increased from 0.07 to 0.10 min −1 and the calculated bile flow from 0.30 to 0.40 ml bile/min. Taken together, these observations illustrate a rapid postprandial recruitment of an apical transport reserve in the healthy liver preventing hepatic bile acid accumulation and securing delivery of bile acids to the intestines.
The present study does not allow for mechanistic conclusions on how the changes relate to potential changes in specific transporter proteins. However, the basolateral transport of 11 C-CSar, mainly facilitated by Na + taurocholate co-transporting polypeptide (NTCP/SLC10A1), 24 did not change in the postprandial state, which is in accordance with a very high capacity of this transport protein, allowing for efficient extraction of bile acids from the sinusoidal blood even under conditions with very high concentration of bile acids returning to the liver. The rapid increase in the apical bile acid transport from hepatocyte to bile, facilitated by the bile salt export pump (BSEP/ABCB11), 25 is most likely mediated by an increased expression of BSEP proteins on the apical membrane. Experimental studies have shown that different substrates such as ursodeoxycholic acid, tauroursodeoxycholic acid, and 4-phenylbutyrate increase the expression of BSEP on the apical membrane, most likely by recruitment from subcanalicular intracellular vesicles. [26][27][28][29][30] This effect is thought to be mediated by a nuclear receptor, farnesoid X receptor (FXR), which can be stimulated directly by bile acids. [31][32][33] FGF-19, which is released to portal blood when bile acids stimulate FXR in the ileocytes, increased in both arterial and hepatic venous blood at the end of the postprandial scan but is likely to have increased in portal blood earlier.
The postprandial intestinal uptake of bile acids may also increase the concentration of bile acids in the portal blood, but because the transit time for bile acids through the small intestines is more than 2 h in healthy humans, there is a rather large window before a significant circulation of 11 C-CSar is expected to be observable. 34,35 Moreover, recirculation of the tracer would primarily be observable in portal venous blood, which we could not sample in the present study. In an invasive pig study performed during fasting, a significant increase in portal venous concentration of 11 C-CSar did not occur until after 75 min. 7 Soloway and Schoenfield, 6 using an invasive setup with bile sampling from a T-tube in the common bile duct following cholecystectomy, observed a postprandial increase in bile flow of 40% and bile salt secretion by 46%. Though their experimental setup interfered with the enterohepatic circulation of bile acids, Soloway and Schoenfield 6 also suggested that the increased biliary secretion of bile acids was caused by increased enterohepatic recirculation rather than by increased de novo synthesis. Our study, which does not interfere with the enterohepatic circulation, validates their findings and interpretations by a direct quantification of the separate transport steps involved in hepatic uptake and secretion of bile acids as well as bile flow.
In conclusion, the transhepatic transport of bile acids increased in the postprandial state compared with that in fasting, and our data show that this was mainly because of an increased apical bile acid secretion. The mechanism is not clear, but we hypothesise that it is driven by increased apical expression of BSEP. Further studies may clarify whether the rapid change could be explained by an increased apical transporter expression as a result of recruitment of internalised vesicular transporters or if other mechanisms are involved.
Financial support