Hepatic and Abdominal Carbon Dioxide Measurements Detect and Distinguish Hepatic Artery Occlusion and Portal Vein Occlusion in Pigs

Authors

  • Soeren Erik Pischke,

    1. Division of Emergencies and Critical Care, Department of Anesthesiology, Oslo, Norway
    2. Intervention Center, Oslo, Norway
    3. Division of Clinical Medicine, Faculty of Medicine, University of Oslo, Oslo, Norway
    Search for more papers by this author
  • Christian Tronstad,

    1. Department of Clinical and Biomedical Engineering, Oslo, Norway
    Search for more papers by this author
  • Lars Holhjem,

    1. Division of Emergencies and Critical Care, Department of Anesthesiology, Oslo, Norway
    Search for more papers by this author
  • Pål Dag Line,

    1. Division of Cancer Medicine, Surgery, and Transplantation, Department of Transplantation Surgery, Oslo University Hospital, Oslo, Norway
    Search for more papers by this author
  • Håkon Haugaa,

    1. Division of Emergencies and Critical Care, Department of Anesthesiology, Oslo, Norway
    2. Division of Clinical Medicine, Faculty of Medicine, University of Oslo, Oslo, Norway
    Search for more papers by this author
    • Soeren Erik Pischke and Håkon Haugaa were supported by unlimited research grants from the South-Eastern Norwegian Health Authority (Helse Sør-Øst).

  • Tor Inge Tønnessen

    Corresponding author
    1. Division of Emergencies and Critical Care, Department of Anesthesiology, Oslo, Norway
    2. Intervention Center, Oslo, Norway
    3. Division of Clinical Medicine, Faculty of Medicine, University of Oslo, Oslo, Norway
    • Rikshospitalet, Oslo University Hospital, P.O. Box 4950, Oslo, Norway 0424. Telephone: +47-230 73692; FAX: +47-230 73681;
    Search for more papers by this author

Abstract

Hepatic artery (HA) occlusion and portal vein (PV) occlusion are the most common vascular complications after liver transplantation with an impact on mortality and retransplantation rates. The detection of severe hypoperfusion may be delayed with currently available diagnostic tools. Hypoperfusion and anaerobically produced lactic acid lead to increases in tissue carbon dioxide. We investigated whether the continuous assessment of the intrahepatic and intra-abdominal partial pressure of carbon dioxide (PCO2) could be used to detect and distinguish HA and PV occlusions in real time. In 13 pigs, the HA and the PV were fully occluded (n = 7) or gradually occluded (n = 6). PCO2 was monitored intrahepatically and between loops of small intestine. The hepatic and intestinal metabolism was assessed with microdialysis and PV as well as hepatic vein blood samples, and the results were compared to clinical parameters for the systemic circulation and blood gas analysis. Total HA occlusion led to significant increases in hepatic PCO2 and lactate, and this was accompanied by significant decreases in the partial pressure of oxygen and glucose. PV occlusion induced a significant increase in intestinal PCO2 (but not hepatic PCO2) along with significant increases in intestinal lactate and glycerol. Gradual HA occlusion and PV occlusion caused steady hepatic and intestinal PCO2 increases, respectively. Systemic clinical parameters such as the blood pressure, heart rate, and cardiac output were affected only by PV occlusion. In conclusion, even gradual HA occlusion affects liver metabolism and can be reliably identified with hepatic PCO2 measurements. Intestinal PCO2 increases only during PV occlusion. A combination of hepatic and intestinal PCO2 measurements can reliably diagnose the affected vessel and depict the severity of the occlusion, and this may emerge as a potential real-time clinical monitoring tool for the postoperative course of liver transplantation and enable early interventions. Liver Transpl, 2012. © 2012 AASLD.

Vascular complications during the course of orthotopic liver transplantation (OLT) lead to liver ischemia, which may constitute a life-threatening state for both the patient and the allograft.1,2 Hepatic artery (HA) stenosis and occlusion are the most common complications (with an incidence up to 12% in adults and as high as 40% in children),1 lead to retransplantation in more than 50% of all cases, and are associated with a high mortality rate.3,4 The incidence of portal vein (PV) obstruction is 1% to 2%1,2 but is higher in patients with splanchnic thrombosis before OLT (up to 15%),5 and the associated morbidity and mortality rates are high.2

The high impact of vascular occlusion on mortality and retransplantation rates underscores the importance of the early detection of vascular complications in the first week after transplantation when most vascular insults occur.3 The current standard of care includes arterial blood lactate assessments, Doppler ultrasound examinations, and daily laboratory liver enzyme assessments.6 However, regular blood lactate measurements are not performed after arterial line withdrawal. Furthermore, there is no consensus on the frequency and timing of Doppler ultrasound examinations; the method does not currently allow continuous monitoring and cannot reliably assess whether a blood flow decrease is detrimental to the graft.3 The release of liver enzymes reflects hepatic cell damage, is thus unspecific with respect to the cause, and increases several hours after vascular occlusion has occurred.3 Accordingly, the detection of severe hypoperfusion may be delayed, and as a result, irreversible graft injury may be inevitable.

Various experimental approaches of continuous surveillance of hepatic metabolism have been evaluated for early detection of liver ischemia. Microdialysis has been successfully used in liver transplant patients7-9 and can detect liver ischemia through increases in lactate levels.8,9 However, hepatic microdialysis has been capable of detecting only HA occlusion (not PV occlusion) in a porcine model10 and clinical studies.9 The continuous measurement of the hepatic partial pressure of oxygen (PO2) and partial pressure of carbon dioxide (PCO2) in animals has been shown to be feasible and has been used to detect liver hypoxia and ischemia in models of global hypoxia11 and hemorrhagic shock.12,13 Likewise, intestinal PCO2 has been shown to reflect intestinal ischemia.13

A tissue PCO2 increase reflects organ hypoperfusion, predicts organ failure, and serves as a measure of both anaerobic metabolism (PCO2 is produced when bicarbonate buffers protons) and blood flow stagnation (the impaired removal of both aerobically and anaerobically produced PCO2).13–16

In a prior study, we evaluated a new miniature PCO2 sensor (IscAlert) in a model of myocardial ischemia/reperfusion and reported that ischemic episodes were successfully identified.17 The results challenged us to consider whether the Food and Drug Administration–approved IscAlert sensor could also be used for the detection of hepatic and intestinal ischemia. Neurotrend sensors and microelectrodes were used to verify IscAlert-obtained values and to obtain tissue PO2 values.

In this study, we aimed to evaluate PCO2 as a real-time diagnostic tool for detecting HA and PV occlusions. We hypothesized that (1) intrahepatic and intraperitoneal PCO2 measurements could enable the distinct detection of HA and PV occlusions and discriminate between them in real time and (2) even gradual HA and PV occlusions could lead to metabolic changes reflected by PCO2 increases in the liver and intestine, respectively.

Abbreviations: CO2, carbon dioxide; HA, hepatic artery; HV, hepatic vein; OLT, orthotopic liver transplantation; PCO2, partial pressure of carbon dioxide; PO2, partial pressure of oxygen; PV, portal vein.

MATERIALS AND METHODS

Experimental Setup

The Norwegian Animal Care and Use Committee approved the following protocol, and the animals were handled according to local guidelines and the Guide for the Care and Use of Laboratory Animals (National Institutes of Health publication 85–23, revised 1996).

Thirteen male and female conventional landrace pigs (Sus scrofa, Noroc) with a mean weight of 59.3 kg (range = 54–62.5 kg) that had no signs of infection were anesthetized with an intramuscular injection of ketamine (30 mg/kg) and azaperone (3 mg/kg). Anesthesia and analgesia were maintained with continuous isoflurane (1%–1.5% inspired fraction) and an infusion of morphine (0.5–1 mg/kg/hour).

After tracheotomy, mechanical ventilation (Servo 900C ventilator system, Siemens, Germany) was used with a mixture of room air and oxygen with the goal of keeping the arterial PO2 value greater than 90 mm Hg (12 kPa). The tidal volume and the ventilation rate were adjusted to keep the arterial PCO2 value close to 40 mm Hg (5.5 kPa).

One catheter was inserted into the right internal jugular vein for intravenous injections and measurements of the central venous pressure, and another catheter was inserted into the right carotid artery for assessments of the arterial blood pressure and arterial blood sampling. The cardiac index was obtained with a PiCCO femoral artery catheter (Pulsion Medical Systems, Munich, Germany) after thermal dilution calibration.

After median laparotomy, the PV and the HA proximal to splitting into lobe arteries were identified in the portal hiatus of the liver. Blood flow measurement with transit time flow probes (PeriVascular probes, MediStim, Oslo, Norway) was established. Inflatable vascular occluders (In Vivo Metric, Healdsburg, CA) were placed around the PV and the HA (Supporting Fig. 1). It was ensured that the deflated occluders did not reduce blood flow, that the maximum inflation totally obliterated the blood flow of the target vessel, and that no other blood vessels were affected. Regulation of the grade of occlusion was achieved with a manually adjustable syringe pump.

Catheters enabling blood sampling from the PV and the hepatic vein (HV) were established with the Seldinger technique through the penetration of the peripancreatic fat and the anterior middle liver lobe, respectively.

Four to 6 miniature sensors (<1 mm in diameter) for measuring PCO2 (IscAlert, SensoCure AS, Horten, Norway) were placed into the liver via a split cannula 5 to 10 mm below the surface (for a demonstration of the IscAlert sensor as well as the insertion and fixation technique, see Supporting Video 1). As many as 3 sensors were freely placed into the abdominal cavity and were lying between loops of small intestine. IscAlert sensors detect changes in PCO2 from an automatically calculated baseline at zero based on a conductometric method described elsewhere.18 Briefly, a gas-permeable membrane allows tissue carbon dioxide (CO2) diffusion into the water-filled sensor, in which CO2 dissolves into HCO3 and H+. This leads to a change in the conductivity of the water, which is continuously measured by the electrical impedance between 2 electrodes in the sensor.

To verify the IscAlert-derived PCO2 changes and to obtain tissue PO2 values, a fiber-optic sensor measuring tissue PO2, PCO2, and temperature values (Neurotrend, Codman, MA) was inserted into the anterior middle lobe of the liver in the same manner as the IscAlert sensors. Neurotrend sensors do not tolerate intestinal movements, which can lead to bending of the sensors' optical fibers. Thus, for 2 animals, 3 dip-type CO2 microelectrodes (Microelectrodes, Inc., Bedford, NH) were placed into the abdominal cavity for comparison with IscAlert-obtained values.

Up to 3 microdialysis catheters (100-kDa cutoff, 3-cm membrane, and 1 μL/minute flow; CMA 65, M Dialysis, Solna, Sweden) were inserted in the proximity of the Neurotrend and IscAlert sensors in the liver, and up to 2 catheters were placed intraluminally into the jejunum. Microdialysis allowed the determination of tissue levels of substances important to cellular metabolism. The microdialysis catheter consisted of a semipermeable membrane with a defined pore size, which allowed the recovery of substances from the tissue. Tissue metabolites equilibrated into a dextran-containing fluid (Plasmodex, Meda AB, Solna, Sweden) on the inner side of the membrane. The fluid was pumped continuously into a collection tube (microvial; M Dialysis). The collection tube was disconnected for the analysis of L-lactate, glucose, pyruvate, and glycerol in an analysis machine (CMA 600, M Dialysis).

After instrumentation, the abdomen was loosely closed with metal surgical clamps to prevent temperature declines and excessive fluid loss due to perspiration but allow intermittent observations of the liver and intestine. The abdominal wall was elevated with a lifting bandage to prevent increased intra-abdominal pressure due to intestinal blood pooling and edema during PV occlusion.

To compensate for fluid loss due to surgery and basic metabolism, crystalloids (0.9% sodium chloride and Ringer's acetate) were administered at 10 to 13 mL/kg/hour to achieve isovolemia, which was determined by the central venous pressure (4–12 cm H2O), cardiac index (<10% change), urine production (>1 mL/kg/hour), and hematocrit (<5% change). Blood glucose was regularly measured, and 10% glucose was infused to ensure normoglycemia. Acute circulatory instability events occurred during the course of PV occlusion, and the pigs were resuscitated with colloid infusions and continuous and bolus administrations of epinephrine.

Experimental Protocol

After a 1-hour postsurgical stabilization period, the subjects were assigned to either group A (7 subjects) or group B (6 subjects). Both experimental setups were designed to produce incremental hepatic injury, maintain baseline conditions as long as possible without the need for additional fluids or vasopressor treatment, and avoid ischemic preconditioning. Accordingly, HA occlusion was carried out before PV occlusion so that a more stable animal model could be achieved. In group A, the HA was fully occluded for 30 minutes, and this was followed by a 30-minute reperfusion period (Fig. 1). Thereafter, the PV was fully occluded for 30 minutes, and this was followed by a 60-minute reperfusion period. Both the HA and the PV were then gradually occluded to 60%, 40%, 30%, 20%, 10%, and 0% blood flow with a duration of 30 minutes at each level (Fig. 1).

Figure 1.

Schematic timeline of the experimental design. The blood flow was reduced in the HA and the PV for the designated intervals in group A (7 animals) and group B (6 animals).

In group B, the baseline HA flow was recorded; thereafter, the flow was gradually reduced to 60%, 40%, 30%, 20%, and 10% of the baseline in 20-minute steps, and it was kept fully occluded for 60 minutes. After a 60-minute reperfusion period, the PV was gradually occluded in the same manner, and 0% flow was maintained for 20 minutes (Fig. 1).

Data Collection

Global variables for circulation and respiration and data collected from Neurotrend sensors were continuously recorded every 20 seconds with ICUpilot software (M Dialysis). IscAlert, microelectrode, and blood flow data were acquired continuously every 20 seconds with LabView version 8.5. Arterial, PV, and HV blood samples were drawn for blood gas analysis (ABL 800, Radiometer, Copenhagen, Denmark) at the baseline, 5 minutes before each new occlusion, and at the end of the last reperfusion period. Microdialysis samples were obtained at the same time points and were analyzed for tissue lactate, glucose, and glycerol (CMA 600, M Dialysis).

In 4 of the 7 animals in group A, IscAlert catheters were used because of a sensor production stop. In 6 animals of the same group, data from Neurotrend sensors were used because of a catheter malfunction in 1 animal. One subject in group A became cardiovascularly unstable after PV occlusion and subsequently died; the data for combined HA and PV occlusions, therefore, consisted of 3 IscAlert data sets. All animals in group B were monitored with both IscAlert and Neurotrend sensors in the liver and intestine.

Arterial blood samples were drawn at the baseline, at the end of each reperfusion period, and at the end of 0% flow in the HA and the PV for analyses of bilirubin, aspartate aminotransferase, alanine aminotransferase, lactate dehydrogenase, alkaline phosphatase, and gamma-glutamyltransferase.

Statistics

Pilot experiments in animals undergoing identical anesthetic procedures, surgeries, and insertions of measurement devices (but without vascular occlusion) revealed no changes in parameters of cardiovascular function or intraorgan metabolism when they were observed as long as the experiment lasted in this study. On the basis of these findings, we decided that a control group would be unnecessary for both ethical and economic reasons. Thus, each animal served as its own control, and values at the end of the stabilization period and before the gradual occlusion of the HA, the PV, or both were used as controls for each parameter in order to monitor biological changes. Maximum values during the experimental events were extracted for all continuously measured parameters (IscAlert, Neurotrend, and systemic parameters). Statistical differences were tested with a linear mixed effects model. A random effects model was used, and we performed model selection for each variable by choosing the model achieving the lowest information criteria. The measured parameters were dependent variables, vascular occlusion events were assigned fixed effects, and the random effects of subjects and groups were included in the model. A post hoc comparison of all groups to the baseline values with Bonferroni correction for multiple testing was performed. A P value ≤ 0.05 was considered statistically significant. Statistical analysis was conducted with SPSS 18 (IBM, Armonk, NY). Graphs were prepared with SigmaPlot 11 (Systat Software, Inc., San Jose, CA) and GraphPad Prism 5 (GraphPad Software, Inc., La Jolla, CA).

RESULTS

Hepatic and Intestinal PCO2 and PO2

In group A, total HA occlusion and PV occlusion were reliably achieved (Fig. 2). HA occlusion led to a significant hepatic PCO2 increase that was detected by both hepatic IscAlert and Neurotrend catheters (P < 0.001 and P < 0.02, respectively; Fig. 2). PCO2 in the abdominal cavity remained unchanged (Fig. 2). During PV occlusion, intestinal PCO2 increased, and this was reliably detected by IscAlert (P = 0.02) and microelectrode sensors (Fig. 2). Hepatic PCO2 increased slowly but nonsignificantly during 30 minutes of PV occlusion (Fig. 2). After the reopening of the PV occlusion, intestinal PCO2 began an immediate return to the baseline, whereas hepatic PCO2 increased before it approached baseline values after 15 minutes.

Figure 2.

HA and PV occlusions are detected by PCO2 measurements. The blood flow in the HA (black line) and the PV (gray line) was reliably reduced to zero during occlusion. HA occlusion (but not PV occlusion) led to significant increases in hepatic PCO2 from the baseline measured by IscAlert and Neurotrend. Both HA occlusion and PV occlusion led to significant reductions in hepatic PO2. Intestinal PCO2 was significantly increased only during PV occlusion (not during HA occlusion) detected by IscAlert sensors. PCO2 and PO2 values are expressed as medians (black lines) and quartile ranges (gray areas). Blood flows are expressed as medians. P ≤ 0.05 (exact values are presented in the text). *Maximum change of 9 mm Hg or 1.2 kPa. †Maximum change of 13 mm Hg or 1.7 kPa. ‡Significant reduction in hepatic PO2. #Maximum change of 43 mm Hg or 5.7 kPa.

During total HA occlusion and total PV occlusion, hepatic PO2 decreased significantly (P < 0.001; Fig. 2). Anoxia was not achieved during HA or PV occlusion.

Hepatic and Intestinal Metabolism

Hepatic microdialysis during the course of HA occlusion revealed a significant lactate increase (P = 0.003), which was accompanied by a significant glucose decrease (P = 0.01; Fig. 3A). Glycerol showed a nonsignificant increasing trend (P = 0.21). During PV occlusion, the hepatic glucose level was unaffected, whereas the lactate level increased significantly (P = 0.002).

Figure 3.

The intermediate meta-bolism in the liver and intestine is affected by HA occlusion and PV occlusion. (A) Hepatic microdialysis showed a significant reduction of glucose and an elevation of lactate during HA occlusion, and this was accompanied by a nonsignificant glycerol increase. PV occlusion led to a significant hepatic lactate increase, but all other parameters were unchanged. (B) HA occlusion did not influence intestinal homeostasis. PV occlusion led to significant increases in intestinal lactate and glycerol. All values are presented as means and standard deviations. *P ≤ 0.05 (exact values are presented in the text).

The intestinal metabolism was not affected by HA occlusion, and this was reflected by unchanged intraluminally assessed microdialysis samples (Fig. 3B). Abrogation of the PV blood flow caused significant increases in intraluminal lactate (P < 0.001) and glycerol (P = 0.004; Fig. 3B).

Blood Gas Analysis and Laboratory Assessment of Hepatic Cell Damage

Blood gas analysis of PV, HV, and arterial blood samples revealed that only PV occlusion and the following reperfusion led to significant and clinically relevant changes in PCO2, lactate, and oxygen saturation at all measurement sites (Table 1).

Table 1. Blood Gas Analysis
 PCO2 (mm Hg)Lactate (mM)Saturation of Oxygen (%)
 ArterialPVHVArterialPVHVArterialPVHV
  • NOTE: All values are presented as means and standard deviations.

  • *

    P ≤ 0.05 (linear mixed effects model with a post hoc comparison of all samples to the baseline and with Bonferroni correction for multiple testing).

Baseline43.4 ± 3.852.6 ± 3.053.7 ± 5.30.70 ± 0.10.84 ± 0.20.51 ± 0.198.3 ± 1.364.2 ± 7.644.6 ± 11.8
HA occlusion43.4 ± 3.052.4 ± 3.055.7 ± 5.30.75 ± 0.20.78 ± 0.20.53 ± 0.197.9 ± 1.259.8 ± 11.734.2 ± 13.1
Reperfusion43.1 ± 3.052.7 ± 2.353.0 ± 3.00.67 ± 0.10.83 ± 0.10.46 ± 0.197.6 ± 1.462.1 ± 8.444.3 ± 14.0
PV occlusion42.3 ± 5.368.9 ± 9.8*60.4 ± 4.52.01 ± 0.83.94 ± 1.7*2.10 ± 1.097.0 ± 2.230.9 ± 17.0*23.5 ± 10.7*
Reperfusion45.2 ± 4.555.5 ± 9.856.6 ± 12.02.66 ± 2.8*2.81 ± 2.82.40 ± 2.9*92.2 ± 7.3*61.9 ± 15.242.7 ± 18.5

A laboratory assessment of the standard clinical parameters for liver cell damage showed a significant aspartate aminotransferase increase from the baseline (40.0 ± 9.5 U/L) during the course of PV occlusion (95.0 ± 52.2 U/L, P = 0.006) and PV reperfusion (123.3 ± 43.3 U/L, P < 0.001), whereas the bilirubin, alanine aminotransferase, lactate dehydrogenase, alkaline phosphatase, and gamma-glutamyltransferase levels remained unaffected and within the normal ranges throughout the experiment.

Systemic Hemodynamics

While complete HA occlusion had no effect, total obstruction of the PV did significantly change the mean arterial blood pressure, heart rate, end-tidal CO2, and cardiac index (P ≤ 0.05; Table 2).

Table 2. Systemic Parameters
EventMean Arterial Pressure (mm Hg)Cardiac Index (L/Minute/m2)Pulse (Beats/Minute)Saturation of Peripheral Oxygen (%)End-Tidal CO2 (mm Hg)
  • NOTE: All values are presented as means and standard deviations for the evaluated period.

  • *

    P ≤ 0.05 (linear mixed effects model with a post hoc comparison of all samples to the baseline and with Bonferroni correction for multiple testing).

Baseline66.8 ± 7.84.2 ± 0.785.3 ± 13.399.7 ± 0.447.3 ± 4.5
HA occlusion70.3 ± 11.34.1 ± 0.885.7 ± 15.399.4 ± 0.947.3 ± 4.5
Reperfusion66.2 ± 8.24.5 ± 0.890.3 ± 18.799.3 ± 0.947.3 ± 3.8
PV occlusion51.2 ± 9.7*3.2 ± 1.0*116.1 ± 29.4*99.6 ± 0.542.0 ± 6.8*
Reperfusion66.5 ± 8.55.1 ± 0.7121.6 ± 11.0*98.1 ± 1.548.0 ± 3.8

Experimental Clinical Situations

An acute total HA or PV obstruction can occur clinically as embolization or thrombosis. Gradual stenosis can also occur because of graft kinking or incomplete thrombosis, which can lead to a total obstruction after some time. Hence, the HA and the PV were gradually occluded separately (Fig. 4A, B) or in combination, with the latter case resembling systemic blood flow impairment seen during cardiogenic or hemorrhagic shock (Fig. 4C). A gradual decrease in the HA blood flow led to a stepwise increase in hepatic PCO2, whereas intra-abdominal PCO2 was unchanged (Fig. 4A). A gradual reduction in the PV blood flow led to an immediate increase in intestinal PCO2 with each reduction in blood flow (Fig. 4B). Combined and graded HA and PV blood flow reduction led to joint increases in hepatic and intestinal PCO2 levels, which reached high endpoint values (Fig. 4C). According to microdialysis, hepatic lactate was unaffected by a gradual reduction of the HA blood flow (data not shown). PV obstructions and combined HA and PV obstructions caused immediate increases in intestinal and hepatic lactate, which attained significance during 10% and 0% blood flow (data not shown).

Figure 4.

Gradual decreases in the HA blood flow, the PV blood flow, or both can be detected by hepatic and intestinal PCO2 measurements with IscAlert sensors. The blood flow in the HA (black lines) and the PV (gray lines) was reliably reduced to target values (expressed as percentages of the baseline blood flow). (A) A gradual decrease in the HA blood flow resulted in a hepatic PCO2 increase (continuous lines), which reached significance with 40% of the baseline blood flow; intraperitoneal PCO2 (dashed-dotted lines) was unaffected. (B) PV occlusion did not affect hepatic PCO2, but intra-abdominal PCO2 started to significantly increase with 20% blood flow. (C) Combined HA and PV flow reduction resulted in significant intestinal and hepatic PCO2 increases with the highest endpoint values for all situations tested. PCO2 values are expressed as medians (black lines) and quartile ranges (gray lines). Blood flows are expressed as medians. P ≤ 0.05 (exact values are presented in the text). *Maximum change from the baseline of 8 mm Hg or 1.1 kPa (hepatic). †Maximum change from the baseline of 39 mm Hg or 5.2 kPa (intestinal). ‡Maximum of 36 mm Hg or 4.8 kPa (hepatic). #Maximum of 50 mm Hg or 6.6 kPa (intestinal).

DISCUSSION

In this study, we have demonstrated that the selective occlusion of the HA and the PV leads to distinct alterations of the hepatic and intestinal metabolism, respectively. HA occlusion affects the hepatic energy metabolism and results in glucose depletion and lactate and PCO2 increases, whereas the intestinal metabolism is unaffected. PV occlusion mainly affects the intestinal metabolism and leads to intraluminal increases in lactate and glycerol that are concomitant with an intra-abdominal PCO2 increase. The gradual occlusion of the HA or the PV alone or the two in combination results in specific PCO2 changes in both the liver and intestine already during states of impaired blood flow.

We detected a significant increase in hepatic PCO2 during HA occlusion and HA stenosis when the blood flow was reduced to 40% of the baseline. Intestinal PCO2 increased only during PV occlusion. The PCO2 increase might have been due to either blood flow stagnation leading to an accumulation of aerobically produced CO2 or anaerobic metabolism leading to CO2 generation by bicarbonate buffering lactate-derived protons.16,19 During complete blood flow obstructions, our study supports the latter cause because simultaneous lactate and glycerol increases indicate ischemia with cellular membrane breakdown. States of diminished blood flow during gradual obstructions did not lead to hepatic or intestinal lactate increases, probably because of preconditioning effects of slow and stepwise reduced blood flow.20 Reduced HA and PV blood flows, however, caused hepatic and intestinal PCO2 increases, respectively, and this strongly indicates that blood flow stagnation contributes to the accumulation of CO2 in the organ.16 PCO2 in the HV showed a modest increase during HA occlusion and PV occlusion. Greater deoxygenation of hemoglobin during hypoxia alkalinizes hemoglobin and thus increases the proton-buffering ability as well as the binding of CO2-forming carboxyhemoglobin (the Haldane effect). This effect enables the transport of large amounts of CO2, whereas dissolved venous effluent PCO2 is only minimally increased.21 Blood gas analysis measures the dissolved portion of CO2, and an increase in tissue PCO2 will, therefore, be more pronounced than an increase in HV PCO2. Tissue PCO2 may thus serve as a parameter of a blood flow assessment, with slower and lower increases during blood flow reductions and fast and higher elevations during states of tissue ischemia.13 Also, systemic vasopressor treatment and profound systemic vasodilatation might lead to organ ischemia22 and could thus be detected with tissue PCO2 measurements. Animal and clinical studies are planned to elucidate this hypothesis.

Standard clinical monitoring using parameters of systemic hemodynamic monitoring and arterial blood gas analysis failed to detect HA occlusion. PV occlusion led to significant changes in the blood pressure, pulse, end-tidal CO2, and cardiac output. These observations are unspecific because they reflect acute hypotension with compromised systemic circulation due to blood pooling in the intestine, and they would thus not result in reliable and exclusive diagnoses of acute PV occlusion. Only aspartate aminotransferase (not alanine aminotransferase or bilirubin) was elevated during PV occlusion. This may imply that intestinal cell damage (but not hepatic cell damage) occurred23 in accordance with aspartate aminotransferase being released during intestinal ischemia. However, the specificity of aspartate aminotransferase as a serum marker for intestinal ischemia is weak.24 The findings support the impression that postoperative HA and PV occlusions are difficult to detect with current standards of monitoring.1 In contrast to this, we have shown that continuous intraorgan monitoring using hepatic and intra-abdominal PCO2 measurements provides both real-time ischemia detection and identification of the affected vascular territory.

HA occlusion and PV occlusion lead to comparable hepatic PO2 decreases because both vessels contribute to the hepatic oxygen content. Therefore, the surveillance of hepatic PO2 does not seem to be optimal for monitoring liver grafts because neither the metabolic state nor the affected vessel can be identified. The occurrence of anaerobic metabolism and significant PCO2 increases seem to be dependent on the lack of HA blood flow and not PV blood flow, and this confirms clinical and experimental findings.9,10 The occurrence of anaerobic metabolism despite sufficient PV blood flow might be explained by the fact that parts of the liver (eg, the biliary system) are dependent on the HA blood flow.4,25

Recently, postoperative monitoring of liver grafts with microdialysis has been successfully evaluated in both experimental and clinical studies and has been shown to detect acute cellular rejection and HA occlusion.8–10 In this study, we have confirmed that HA occlusion already leads to significant lactate production after 30 minutes, and this is accompanied by PO2 decreases and PCO2 increases; this indicates that lactate production is indeed due to anaerobic metabolism rather than hypermetabolism as seen in rejection.10 PV occlusion affects the intestinal metabolism, and it has been shown that intestinal ischemia leads to dysfunction of the gut barrier, which results in luminal lactate and glycerol increases.26,27 Microdialysis catheters placed between loops of intestine have been shown to reflect the intraluminal changes.27,28 Results from this study might, therefore, be translated to clinical monitoring when microdialysis catheters or IscAlert sensors would be placed between intestinal loops for the reliable detection of intestinal ischemia.

PV occlusion led to significant hepatic lactate increases, and this was not observed in a different model in which a portocaval shunt was used during PV occlusion.10 The discrepancy can be explained as follows: the portocaval shunt prevented intestinal venous congestion, whereas in our model, blood stasis and pooling in the intestine led to intestinal ischemia with lactate production and assumed shunting to the systemic circulation. In addition to physiological and, therefore, normal shunts, patients with portal hypertension have established collaterals before transplantation.29 However, these collaterals are probably not as efficient as an experimental portocaval shunt during postoperative PV occlusion. Thus, PV occlusion after OLT will lead to some degree of intestinal venous congestion with subsequent intestinal ischemia, lactate and PCO2 increases.30

After liver transplantation, the most important property of a test is its ability to detect impaired HA circulation because this is the most deleterious and common vascular complication. Consequently, both microdialysis and PCO2 monitoring could be useful for detecting and discriminating between HA occlusion and PV occlusion. Currently, microdialysis has the advantage of clinical approval, at least in Europe. However, the additional workload of substance analysis and system maintenance, which requires trained personnel, in combination with the considerable price and the uncertainty of individual catheter life spans, may limit future clinical applications in liver transplantation. IscAlert PCO2 sensors, on the other hand, would provide the advantage of real-time, continuous monitoring at the bedside without an increased workload at an envisioned lower price. Most importantly, according to our results, partially decreased blood flows would be detected only with IscAlert sensors, and full occlusion would be detected earlier with IscAlert sensors versus microdialysis catheters. Both IscAlert and microdialysis catheters are inserted into the transplant via a split-cannula technique (Supporting Video 1), and no complications such as bleeding or infection have been observed in a clinical study of 73 patients using microdialysis catheters.9

The limitations of this study include the large variances in systemic circulatory function during PV occlusion. However, because the trends of all measured variables were consistent, statistical significance was achieved. The number of IscAlert sensors differed between organs, and especially in the intestine, fewer sensors than desired were applied. This was due to a transient stop in IscAlert sensor production and the prioritization of PCO2 assessment in the liver versus the intestine. We did not evaluate hepatic and intestinal PCO2 during HV occlusion, which is a rare clinical condition caused by veno-occlusive disease or caval stenosis.1,31 One might speculate that total HV occlusion would lead to a PCO2 rise in both the liver and the intestine because of blood congestion and tissue ischemia, but further studies should explore this disease entity. Currently, none of the investigated PCO2 sensors are approved for hepatic or intestinal use. Neurotrend sensors neither are produced nor are commercially available, whereas microelectrodes are not approved for clinical use. The IscAlert sensor is Food and Drug Administration–approved for use in human skeletal muscle, and approval for intra-abdominal use will be sought. The IscAlert sensor design has good functionality: the sensor was easy to insert into both the liver and the abdominal cavity, did not lead to bleeding or other adverse effects, and was easy to retract without damage to the liver tissue at the end of the experiment.

Clinical studies will be conducted with 2 IscAlert PCO2 sensors inserted into the liver at the end of transplantation surgery (one in the left lobe and one in the right lobe). One to 2 intestinal sensors could either be left in the abdominal cavity at the end of surgery or be inserted percutaneously under ultrasound guidance via a split-cannula technique.28

In conclusion, continuous, real-time intraorgan monitoring of liver and intestinal PCO2 levels evolves as a potential method for monitoring the hepatic and intestinal blood supply as long as transplant animal studies and clinical studies in OLT patients confirm the presented results. Vascular complications can be revealed almost immediately, and this is clearly superior to current monitoring methods. Combined hepatic and intra-abdominal sensor placement enables the detection and discrimination of HA and PV occlusions.

Acknowledgements

The nursing staff of the Intervention Center provided excellent assistance during the experiments.

Ancillary