The experiments were approved by the regional animal ethics committee and were conducted in accordance with the directive of the European Union for the protection of animals used for scientific purposes [24]. In this study, 24 healthy 3-month-old domestic pigs (a crossbreed between Swedish country breed, Hampshire and Yorkshire) of both sexes and with a median body weight of 30 kg (range 25–41 kg) were used.

On the farm, the animals were given azaperone (200 mg, i.m.; Elanco, Herlev, Denmark). On arrival at the research facility, anesthesia was induced by tiletamine (6 mg kg⁻¹, i.m.; Virbac, Kolding, Denmark), zolazepam (6 mg kg⁻¹, i.m.; Virbac), and azaperone i.m. (4 mg kg⁻¹). Atropine (1.5 mg, i.m.; Mylan, Stockholm, Sweden) was given. Two peripheral catheters (1.1 mm, Venflon™ Pro Safety, BD, Helsingborg, Sweden) were inserted in auricular veins. Propofol (1–2 mg kg⁻¹, i.v.; Fresenius Kabi, Uppsala, Sweden) was given if needed. The pigs were orally intubated with a 6-mm endotracheal tube (Covidien, Tullamore, Ireland). Anesthesia was maintained with propofol (10–15 mg kg⁻¹ h⁻¹, i.v.) and fentanyl (5–20 μg kg⁻¹ h⁻¹, i.v; Meda, Solna, Sweden) given by motorized syringe pumps (Alaris CC, Cardinal Health, Rulle, Switzerland). The depth of anesthesia was intermittently checked by pain response. Cefuroxim (750 mg, i.v.; GSK, Solna, Sweden) was given before, and heparin (5000 IU, i.v.; LEO Pharma, Malmö, Sweden) after instrumentation. Ringer’s acetate (10 ml kg⁻¹ h⁻¹, i.v.; Fresenius Kabi) and 10% glucose with 40 mM sodium and 20 mM potassium (0.5 ml kg⁻¹ h⁻¹, i.v.; Fresenius Kabi) were administered, by volume pumps (Alaris GP, CareFusion). The pigs were ventilated in volume control ventilation mode (PV 501, Breas Medical AB, Sweden) to achieve arterial PCO₂ of 4.8–5.5 kPa, and the fraction of inspired O₂ was adjusted to maintain arterial PO₂ at 11–18 kPa. The positive-end expiratory pressure was set at 4–6 cm H₂O. Respiratory variables and gases were measured at the endotracheal tube (AS/3, Datex, Helsinki, Finland). The body temperature was kept at 37.5–38.5 °C by using a thermal mattress and a forced-air warming blanket. At the end of the experiment, the anesthetized animals were euthanized by a rapid i.v. injection of 40 mmol potassium chloride (B. Braun, Danderyd, Sweden), and asystole and circulatory arrest were confirmed by ECG and blood pressure recordings.

In the left external jugular vein, a 7-Fr triple-lumen central line (Arrow/Vingmed, Järfälla, Sweden) was inserted for fluid and drug administration. In the right carotid artery, a 4-Fr introducer (Cordis Corporation, Miami Lakes, FL, USA) was positioned for measurement of systemic blood pressure and heart rate and also for blood sampling. In the right external jugular vein, an 8-Fr introducer was inserted (Cordis Corporation). Through this, a pulmonary artery catheter (Swan-Ganz CCOmbo, 7.5 Fr, Edwards Lifesciences, Irvine, CA, USA) was positioned, with its tip in a branch of the pulmonary artery, under the guidance of the pressure curve. In addition, correct placement was confirmed by fluoroscopy (Philips BV 300, Stockholm, Sweden). The pulmonary artery catheter was used for blood sampling and measurements of pulmonary capillary wedge pressure (PCWP), central venous pressure (CVP), and CO using the semi-continuous thermodilution technique (Vigilance, Edwards Lifesciences).

A midline abdominal incision was performed. A 14-Fr Foley catheter (Unomedical, Flintshire, UK) was inserted in the bladder and fixed with a purse string suture. A laser Doppler probe (Probe 427, Perimed, Järfälla, Sweden), connected to a laser Doppler device (Periflux System 5000, Perimed), was placed intraluminally, pointing toward the intestinal mucosa in a mid-jejunal intestinal loop approximately 30 cm distal to the Treitz ligament for estimation of local mucosal perfusion [25]. An ultrasonic transit-time flowmeter (Vascular TTFM Probe 6 mm, Medistim ASA, Oslo, Norway) was placed around the superior mesenteric artery (SMA). A 6-Fr catheter (Nutrisafe 2, Vygon, Ecouen, France) was inserted in a distal branch and advanced to the superior mesenteric vein for pressure measurement and blood sampling. The blood samples were analyzed for pH, blood gases, electrolytes, lactate, hemoglobin, and glucose, using either i-STAT (Abbott Scandinavia, Solna, Sweden) or GEM Premier 4000 (Instrumentation Laboratory, Lexington, MA, USA), at 37 °C. Blood pressures were measured by a pressure transducer (Codan, Forstinnigen, Germany) connected to the AS/3 (Datex). A free-floating microdialysis catheter (62 Gastrointestinal Microdialysis Catheter, membrane length 30 mm, cut-off 20 kDa, M Dialysis, Stockholm, Sweden) was placed intraperitoneally in the left lower quadrant [26]. A syringe pump (107 Microdialysis Pump, M Dialysis) was used to propel a solution (Perfusion fluid T1, M Dialysis) at 2 μL min⁻¹ through the microdialysis catheter. The microdialysate was analyzed for glucose, glycerol, lactate, and pyruvate (CMA 600, M Dialysis).

In the control and the tamponade groups, a 12-Fr Foley catheter (Unomedical) was inserted into the pericardial space via a small incision through a diaphragmatic window and sutured in place. Subcutaneous fat was used as pledgets to reduce the risk of leakage. Approximately 5 ml of physiological sodium chloride solution (B. Braun) was instilled in the balloon of the Foley catheter.

In the animals allocated to the caval vein balloon group, an additional 11-Fr introducer (Cordis Corporation) was inserted in the right external jugular vein. A balloon catheter (Cordis PTA Dilatation Catheter, large diameter balloon, 7 Fr, 80 cm, Cordis Corporation) was advanced to the diaphragmatic part of the inferior caval vein under fluoroscopic guidance.

After an intervention-free period of 1 h to achieve stable baseline conditions, the animals were divided into three groups: (1) the tamponade group, (2) the caval vein balloon group, and (3) the control group. In the tamponade group, CO was reduced by instillation of a colloid fluid (at 38 °C; 6% hydroxyethyl starch, Fresenius Kabi) via the pericardial Foley catheter into the closed pericardium. Cardiac tamponade was confirmed by transthoracic echocardiography (SonoSite Titan™, L52/10–5 MHz Transducer, Askim, Sweden). In the caval vein balloon group, CO was decreased by partial inflation of the balloon. Inflation of the balloon was confirmed by fluoroscopy. After collecting baseline values, blood samples, and microdialysate, CO was stepwise lowered in these groups to 75% (CO_75%), 50% (CO_50%), and 35% (CO_35%) of baseline value for 1 h each, at which point respiratory and hemodynamic variables as well as blood samples were collected. The microdialysate was gathered during the last 30 min at each CO level. In the control group, no intervention was made to affect CO, but data collection was identical to that of the two other groups.

Cardiac output and SMA blood flow were divided by body surface area, according to the formula: body surface area (m²) = 0.0734 × weight (kg)^0.656, to obtain the cardiac index (CI) and the SMA blood flow index [27]. Oxygen saturation was calculated as PO₂^2.94/(PO₂^2.94 + P₅₀^2.94). 4.76 kPa was used as the porcine partial pressure of oxygen (PO₂), where hemoglobin is half-saturated (P₅₀), and the value was adjusted with the fixed acid Bohr coefficient [28]. Oxygen content was calculated as saturation (fraction) × Hb (g/l) × 1.34 + 0.225 × PO₂ (kPa) [29]. Oxygen extraction was calculated as arterial oxygen content subtracted by venous oxygen content. Oxygen delivery (DO₂) was acquired as the blood flow index times arterial oxygen content, and oxygen uptake (VO₂) was calculated as oxygen extraction times the blood flow index.

Data are presented as medians with interquartile ranges. In the statistical analysis, the absolute value was used at baseline, and thereafter, the difference or percentage change from baseline was used. The Bonferroni corrected Kruskal-Wallis tests were performed to identify statistical differences between groups at each CO level, and, if significant, they were followed by pairwise comparisons between groups using the Bonferroni corrected Mann-Whitney U tests. Least-squares linear and logarithmic regressions were made. A P value of less than 0.05 was regarded as statistically significant. Statistical analysis was performed using IBM SPSS Statistics version 22.0 for Windows (SPSS Inc., Chicago, IL, USA).

Of the 24 animals used in the study, one pig was excluded due to circulatory instability after instrumentation, and one died suddenly during instrumentation. Nine animals were allocated to the tamponade group; two were excluded because of leakage from the pericardial sutures that could not be repaired, and one was excluded due to failure to collect microdialysate. Four of the six animals in the tamponade group died at CO_35% and were only included in the statistical analyses until CO_50% (Fig. 1b). Seven pigs were allocated to the caval vein balloon group; one died of an arrhythmia at CO_75% and was excluded. Six pigs were included in the control group.

At baseline, CI values were 6.4 (5.5–7.5), 5.9 (4.8–6.6), and 6.7 (6.0–7.5) l min⁻¹ m⁻² in the control (n = 6), tamponade (n = 6), and caval vein balloon (n = 6) groups, respectively. Both inflation of the caval vein balloon and induction of the cardiac tamponade caused a controllable and steady lowering of CO to the targeted level, whereas the controls maintained their CO level throughout the study (Fig. 1a). After the desired level of CO was reached, no major adjustments of the pericardial volume or caval vein balloon inflation were required (data not presented).

At CO_75%, the mean arterial pressure (MAP) was unchanged in the tamponade and caval vein groups, compared to the control group (Fig. 2a). In both intervention groups, MAP decreased significantly when CO was further reduced (P < 0.05 compared to the control group, Fig. 2a). Heart rate followed a biphasic response in both intervention groups (Fig. 2b). CVP increased in the tamponade group and decreased in the caval vein group (P < 0.05 in both groups compared to the control group), while the mesenteric venous pressure (MVP) tended to increase in both intervention groups (Fig. 2c, d).

At baseline, SMA flow index values were 1.03 (0.90–1.42), 0.85 (0.60–1.29) and 1.04 (0.92–1.40) l min⁻¹ m⁻² in the control, tamponade, and caval vein balloon groups, respectively. SMA blood flow decreased significantly in both intervention groups as CI decreased (P < 0.05 compared to the control group, Fig. 2e). Intestinal mucosal perfusion was decreased at CO_35% (P < 0.05 for caval vein balloon versus controls, Fig. 2f).

The decrease in SMA flow in the intervention groups followed the decrease in CI, but the SMA flow was higher than predicted at CO_50% (61% [48–73] and 60% [53–70]) in the tamponade and caval vein groups, respectively) and lower at CO_35% (17% and 24% [19‐36] in the tamponade and caval vein groups, respectively; Figs. 2e and 3). To further explore the relationship between CI and SMA flow, linear and logarithmic regressions were carried out. Using SMA flow (percentage of baseline) as the dependent variable and CI (percentage of baseline) as the independent variable, regression analysis resulted in a better fit when using a natural logarithmic rather than a linear equation (Fig. 3).

Arterial PO₂ and PCO₂ and also minute ventilation and fraction of inspired O₂ were normal, stable, and similar in all groups throughout the study (Table 1).

At baseline, systemic DO₂ values were 550 (490–690), 690 (590–820), and 730 (590–810) ml O₂ min⁻¹ m⁻² in the control, tamponade, and caval vein balloon groups, respectively, while mesenteric DO₂ were 92 (75–138), 108 (62–170), and 116 (103–131) ml O₂ min⁻¹ m⁻² in the respective groups. Systemic respective mesenteric VO₂ were 300 (260–300) and 38 (27–59) ml O₂ min⁻¹ m⁻² in the control group, 350 (250–430) and 47 (34–61) ml O₂ min⁻¹ m⁻² in the tamponade group, and 360 (330–390) and 50 (46–68) ml O₂ min⁻¹ m⁻² in the caval vein balloon group. Systemic and mesenteric DO₂ decreased significantly when CO was reduced (P < 0.05 compared to control, Fig. 4a, b). In parallel, the mixed venous and mesenteric venous SO₂ decreased significantly to very low values at CO_35% (P < 0.05 compared to controls, Fig. 4c, d). Significant decrements in systemic and mesenteric VO₂ developed at CO_35% (P < 0.05, compared to controls, Fig. 4e, f).

Arterial pH decreased in the intervention groups and was significantly lower compared to the control group (P < 0.05) at CO_50% and CO_35% (Table 1). Arterial and mesenteric lactate levels increased significantly from CO_75% and CO_50% in the tamponade and caval vein balloon groups, respectively (P < 0.05 compared to the control group, Fig. 4g, h). The arterial-mesenteric venous differences of lactate tended to become more negative as CO was progressively reduced (Table 1). Intraperitoneal lactate concentrations increased significantly in both intervention groups and were significantly higher compared to the control group from CO_50% (P < 0.05, Table 1). Intraperitoneal lactate/pyruvate ratios increased from CO_50% and CO_35% in the tamponade and caval vein balloon groups, respectively (P < 0.05 compared to the control group, Table 1). Intraperitoneal glycerol levels increased at CO_35% in the caval vein group (P < 0.05 compared to the control group, Table 1). Intraperitoneal pyruvate and glucose concentrations and plasma glucose concentrations did not change significantly (Table 1).