The myocardial oxygen consumption (MV˙o 2) to left ventricular pressure-volume area (PVA) relationship is assumed unaltered by substrates, despite varying phosphate-to-oxygen ratios and possible excess MV˙o 2 associated with fatty acid consumption. The validity of this assumption was tested in vivo. Left ventricular volumes and pressures were assessed with a combined conductance-pressure catheter in eight anesthetized pigs. MV˙o 2 was calculated from coronary flow and arterial-coronary sinus O2 differences. Metabolism was altered by glucose-insulin-potassium (GIK) or Intralipid-heparin (IH) infusions in random order and monitored with [14C]glucose and [3H]oleate tracers. Profound shifts in glucose and fatty acid oxidation were observed. Contractility, coronary flow, and slope of the MV˙o 2-PVA relationship were unchanged during GIK and IH infusions. MV˙o 2 at zero PVA (unloaded MV˙o 2) was 0.16 ± 0.13 J ⋅ beat− 1 ⋅ 100 g− 1 higher during IH compared with GIK infusion (P = 0.001), a 48% increase. The study demonstrates a marked energetic advantage of glucose oxidation in the myocardium, profoundly affecting the MV˙o 2-PVA relationship. This may in part explain the “oxygen-wasting” effect of lipid-enhancing interventions such as adrenergic drugs and ischemia.
- left ventricle
- cardiac energetics
- pressure-volume area
suga introduced (34) the myocardial oxygen consumption (MV˙o 2) to pressure-volume area (PVA) relationship in 1979 as an effective model to describe left ventricular energetics (Fig. 1). Since then it has been used to characterize the effects of numerous interventions on cardiac mechanoenergetic function (35), and the MV˙o 2-PVA model has been increasingly used in intact animal and human studies (13, 23, 36). However, extrapolating the model from the well-controlled isolated heart preparation (34) to the in vivo clinical situation might have inherent problems concerning physiological variations in substrate availability. The unloaded MV˙o 2 in the MV˙o 2-PVA model has been assumed unaltered by myocardial substrate metabolism (35), despite the well-known increase in MV˙o 2induced by fatty acids (22, 38). This assumption was based on the low variation in phosphate-to-oxygen ratios (P:O) for myocardial substrates, which suggests a maximum 11% higher unloaded MV˙o 2 for isolated free fatty acid (FFA) consumption compared with glucose (35). An increased supply of FFA will diminish the rate of glucose oxidation to a minimum and approach this theoretical limit (25). Additionally, high levels of FFA might increase oxygen consumption further, both by uncoupling of oxidative phosphorylation (3) and by inducing cycling of FFA in and out of the triglyceride pool, a futile, energy-consuming cycle (24, 25). Finally, previous studies indicate that high levels of FFA can influence energy consumption related to excitation-contraction (EC) coupling (5) and other related Ca2+-handling processes in myocytes (14, 26).
The aim of the present study was to assess whether altering myocardial substrate availability in an in vivo situation with undisturbed myocardial perfusion would alter the unloaded MV˙o 2 of the MV˙o 2-PVA relationship. Because cardioactive drugs and disease states may alter the circulatory levels of myocardial substrates (25), the metabolic contribution to variations in cardiac energetics needs to be quantified.
MATERIALS AND METHODS
The experimental protocol was approved by the local steering committee of the Norwegian Experimental Animal Board. All studies were conducted in compliance with institutional animal care guidelines, the National Institute of Health's Guide for the Care and Use of Laboratory Animals [DHHS Publication No. (NIH) 85–23, Revised 1985], and the “Guiding Principles in the Care and Use of Laboratory Animals” of the American Physiological Society.
Eight castrated male pigs (Sus scrofa domesticus, Norwegian strain, 23–32 kg) were used. They were fasted overnight with free access to water. Intramuscular ketamine (20 mg/kg) and atropine (1 mg) were used as premedication, and pentobarbital (10 mg/kg) and fentanyl (0.01 mg/kg) boluses were used as anesthetic induction intravenously. The pigs were tracheostomized and ventilated with an air-oxygen mixture on a volume-controlled respirator ( = 0.5, Servo 900, Elema-Schönander). Continuous intravenous anesthesia was administered through a central venous catheter (CVC): pentobarbital sodium (4 mg ⋅ kg− 1 ⋅ h− 1), fentanyl (0.02 mg ⋅ kg− 1 ⋅ h− 1), and midazolam (0.3 mg ⋅ kg− 1 ⋅ h− 1). Mean arterial pressure (MAP) and central venous pressure (CVP) were measured in the thoracic aorta and right atrium, respectively. An additional catheter was introduced to the abdominal aorta for arterial blood sampling. A sternotomy was performed, and the left hemiazygos vein was ligated at its passage through the pericardium. Transit-time ultrasonic flow probes were placed proximal on the left anterior descending and circumflex coronary arteries and on the root of the pulmonary artery (CardioMed CM-4000, Medi-Stim AS). Hemodynamic variables were continuously sampled, digitized, and stored (LabView 3.1.1, National Instruments). The coronary sinus was catheterized through the left thoracic wall via the ligated left hemiazygos vein. A 7-Fr balloon catheter (Sorin Biomedical) for caval occlusion was advanced to the proximal part of the caudal vena cava. A 6-Fr, 12 electrode, dual-field, pigtail-combined microtip and conductance catheter (Millar Instruments) was positioned along the left ventricular long axis and connected to a conductance conditioner (Leycom Sigma-5, Cardiodynamics BV). Pressure calibration was performed before insertion. Positioning of the catheter was evaluated by palpation of the apex. Blood volume was maintained by intravenous infusion of 0.9% NaCl. Urine was drained through a cystostoma.
Figure 2 outlines the different sequences of the protocol. A crossover design for administration of glucose-insulin-potassium (GIK) and Intralipid-heparin (IH) solutions was used. After the experimental preparation and a stabilization period of 15 min, baseline values were measured at T 1. AtT 2 and T 4, two consecutive rapid total vena caval occlusions (VCO) were recorded for assessment of contractility. During intervalsT 2–T 3 andT 4–T 5, seven steady-state data acquisitions were performed, including MV˙o 2 and pressure-volume data, starting at uninfluenced preload and continuing through subsequent steps of preload reduction (MAP decrement of 5–10 mmHg). Lowest recorded MAP level was 49 ± 5 mmHg. A period of 60–90 s was needed to reach a desired steady-state MAP level by partly occluding the caval lumen. Pressure-volume data were then recorded over a period of 10 s, while simultaneously a blood sample for oxygen saturation measurement was drawn from the coronary sinus and coronary blood flow was assessed. Respiratory influence on hemodynamics was avoided by disconnecting the respirator during pressure-volume sampling. The half-time of coronary flow adaptation to decreased perfusion pressure was assumed ∼5 s, as described earlier (6). Arterial Hb and oxygen saturation were assessed at uninfluenced preload.
At the end of the experimental protocol, hearts were stopped by intraventricular injection of 20 ml KCl (1 mmol/ml). The hearts were excised for Evans blue (1%) staining of the perfusion area of the left coronary artery. The weight of stained heart muscle (87 ± 15 g) was used for correction of indexes to 100 g wet wt.
GIK and IH solutions.
GIK and IH solutions were prepared before each experiment. The GIK solution contained 60 g d-glucose in 100 ml 0.9% NaCl with 10.2 U insulin (Actrapid, Novo Nordisk) and 10.2 mmol KCl to a total volume of 148 ml, osmolarity at ∼2,600 mosmol/l. GIK was administered at 2.6 ml ⋅ kg− 1 ⋅ h− 1(CVC). Intralipid (200 mg/ml, Pharmacia) is made of 200 mg/ml purified soybean oil, emulsified with purified egg-phospholipids, and adjusted osmotically with glycerol; osmolality at ∼350 mosmol/kgH2O (ref. Pharmacia). The IH solution contained 100 ml Intralipid with 11,000 U of heparin Na (Nycomed Pharma) and was administered at 3 ml ⋅ kg− 1 ⋅ h− 1(CVC). Heparin Na was used for activation of lipoprotein lipase.
Determination of substrate oxidation.
Infusion of isotopes was started 45–60 min before baseline (T 1) with a bolus of 30 ml/h for 15 min, continued by steady-state infusion at 8 ml/h throughout the experiment. We used [9,10-3H(N)]oleic acid (cat. no. NET289) andd-[14C(U)]glucose (NEC042X) (both NEN Research Products, Du Pont, Germany), dissolved in porcine plasma to a final radioactivity of 25 and 7 μCi/ml (oleate and glucose, respectively). At each time point (T 1–T 5), arterial and sinus coronarius blood samples were drawn simultaneously. Three 1-ml samples for 14CO2 determination were transferred to airtight 14CO2 trapping vials (39). Four 1.25-ml aliquots were cool-centrifuged, and the plasma was immediately frozen on liquid N2. Plasma was stored at −70°C and analyzed later for determination of3H2O and substrate levels. The content of3H2O in plasma was determined by vacuum sublimation as described by Midwood (21). The14CO2 content of the blood was assessed by a diffusion method as described by Wisneski et al. (39). Trapped14CO2 and plasma-3H2O were counted on a β-scintillation counter (Packard 1900 TR Liquid Scintillation Analyzer, Packard Instruments BV). Glucose and FFA oxidation rates (μmol ⋅ min− 1 ⋅ 100 g− 1) were then calculated at each time point according to Equation 1 where (count)Δ is the coronary sinus-to-arterial difference in end products (14CO2 or 3H, dpm/ml), SA is the specific activity of the radioactive substrates in plasma, and CF is coronary blood flow (ml ⋅ min− 1 ⋅ 100 g− 1) (17). The specific activities of substrates were as follows: for glucose, 3.6 ± 1.1 vs. 1.4 ± 0.4 vs. 3.4 ± 1.3 dpm/nmol; and for FFA, 44 ± 25 vs. 47 ± 14 vs. 6.4 ± 2.2 dpm/nmol (at baseline vs. GIK vs. IH, respectively).
Hb oxygen saturation was measured on a hemoximeter (OSM2 Hemoximeter, Radiometer). Blood gases were analyzed on a standard blood gas lab (BGM 1312, Allied Instrumentation Laboratory). Hb was analyzed from EDTA-blood on a cell analyzer (CA 460, Medonic). Plasma levels of glucose, lactate, and FFA were determined on a centrifugal analyzer (Cobas Fara II, Roche Diagnostica; kits and quality controls from Boehringer Mannheim).
The conductance-catheter method has been extensively described earlier with respect to technical and methodological aspects (2, 15), accuracy (1), and limitations (4). We used the dual-field technique (33). Calculations were done using the Conduct-PC software (CPCW version V3.15, Cardiodynamics BV). The heart cycle was defined to start at the peak of the R wave in the QRS complex, corresponding to end diastole. End systole is calculated as proposed by Sagawa (29). Because of uncertainties related to determining parallel volume (Vp) (37), representing nonblood conductance, and the gain correction factor (α) (37), these corrections were not used because we were interested in the relative volume changes only.
Myocardial oxygen consumption (J ⋅ beat− 1 ⋅ 100 g− 1) was calculated as Equation 2 where CF is coronary blood flow (ml/min), (O2-sat)Δis the difference between arterial and coronary sinus O2saturations (%), Hb is hemoglobin (g/ml), 1.39 is a constant for ml O2/g Hb, and HR is heart rate (beats/min).
PVA (J ⋅ beat− 1 ⋅ 100 g− 1) was calculated as (40) Equation 3where SW is stroke work, calculated as an integral of all sampled pressures and volumes between end diastole and end systole (mmHg ⋅ ml). Pes is end-systolic pressure (mmHg), Ves is end-systolic volume (ml), V0 is volume-axis intercept of the linear end-systolic pressure volume relationship (ESPVR) from consecutive beats during rapid total VCO, Ped is end-diastolic pressure (mmHg), and Vedis end-diastolic volume (ml).
We utilized two indexes reflecting left ventricular contractility: the slope (PRSWI) of the linear SW-Ved relation (preload recruitable stroke work) (11), and the slope (E es) of the linear ESPVR (35). PRSWI and E es were assessed twice at times T 2 andT 4 from consecutive beats during rapid total VCO, where a mean value was used at each time point.
Sampled and calculated variables were sorted according to treatment (baseline, GIK, or IH), where the effect of treatment was tested with two-way ANOVA so that any systematic difference between experiments was removed (20). A post hoc test (Tukey's) for multiple comparisons was used where appropriate. Paired t-test was used for comparison of isolated GIK and IH effects. Assessment of the effect of GIK and IH infusions on the total pool of MV˙o 2 and PVA values was evaluated with analysis of covariance, in a multiple linear regression model with a dummy variable coding for GIK and IH infusion (16). Values are reported as means ± SD. Significance was accepted at P< 0.05. Calculations and statistics were performed using a spreadsheet and a statistical package (Microsoft Excel 7.0 and SPSS 8.0.0).
Table 1 shows the general hemodynamic variables at control preload. The MAP was lower during GIK compared with IH infusion. This is also reflected in a higher Pesduring IH infusion compared with GIK as seen in Table2. A summary of left ventricular volumes at uninfluenced preload is given in Table 2. Indexes of left ventricular contractility are presented in Fig. 3, and both PRSWI and E es were unchanged during alternating substrate infusions. Rate of pressure development, dP/dt max, was unchanged during infusions, 2,000 ± 300 vs. 2,150 ± 600 mmHg/s (IH vs. GIK). Rate of diastolic relaxation, dP/dt min, was slightly augmented by IH compared with GIK infusion (−2,250 ± 200 vs. −2,100 ± 300 mmHg/s; P < 0.05). The V0 used for PVA calculation was 58 ± 18 ml for the GIK condition and 65 ± 22 ml for the IH condition.
Substrate uptake and oxidation.
At baseline (T 1) the arterial plasma values of myocardial substrates were as follows (mmol/l): 5.7 ± 0.6 glucose, 0.41 ± 0.28 FFA, and 1.3 ± 0.6 lactate. During GIK infusion the arterial glucose level increased to 10.8 ± 2.0 mmol/l, whereas FFA levels increased to 3.1 ± 1.0 mmol/l during IH infusion. Independent of crossover order, glucose levels were unchanged from baseline during IH infusion, as were FFA levels during GIK infusion. During experiments, lactate levels were never measured higher than 2.0 mmol/l. Myocardial uptake of substrates and calculated oxidation rates are presented in Fig. 4. The uptake of glucose increased (P < 0.05) and lactate tended to increase (P = 0.08) compared with baseline during GIK infusion, whereas both fell below baseline during IH infusion. The opposite pattern was shown for the uptake of FFA. Glucose oxidation was significantly increased during GIK infusion, as was the FFA oxidation during IH infusion. The baseline condition is comparable with an overnight fasting state, where ≥30% of the ATP production in sum is derived from substrates not marked in the present study (lactate, pyruvate, triglycerides, and ketone bodies) (25). ATP production, from the glucose and FFA oxidation rates, was predicted using an ATP-per-mole substrate ratio of 32:1 for glucose and 105:1 for FFA (25). The ratio between FFA-derived and glucose-derived ATP, accounted for by the calculated oxidation rates, was 2.19:1 at baseline, 0.26:1 in the GIK condition, and 23.5:1 in the IH condition (P < 0.001).
Oxygen consumption and MV˙o2-PVA relations.
Hb values were unchanged throughout the protocol (8.1 ± 0.4 g/dl atT 2 and 7.5 ± 1.2 g/dl at T 4) and unchanged also when sorted to GIK and IH groups. At uninfluenced preload, during GIK versus IH infusions, there was no significant difference in the oxygen extraction: 68 ± 8% vs. 68 ± 6%, CF (Table 1) and MV˙o 2; 7.6 ± 1.2 vs. 8.3 ± 1.5 ml O2 ⋅ min−1 ⋅ 100 g− 1 (P = 0.08). Table3 presents the relationships between MV˙o 2 and PVA derived during GIK and IH conditions. Slopes were unchanged from substrate infusions. A significantly higher y-intercept (unloaded MV˙o 2) was found during IH infusion compared with GIK. This was further explored using analysis of covariance performed on the total pool of MV˙o 2 and PVA data. As seen from Fig. 5, a significant mean difference in MV˙o 2 between GIK and IH groups was found as well as a high probability of parallel relationships (no interaction). At zero PVA, this amounts to a 48% higher unloaded MV˙o 2 during IH compared with GIK infusion.
The important result of this study is the profound influence of substrate metabolism on the unloaded MV˙o 2 of the PVA-MV˙o 2 relation in an intact animal model. The efficiency of the left ventricle (mechanical energy output related to MV˙o 2) was reduced during increased rates of fatty acid metabolism, and this metabolic influence on the MV˙o 2-PVA relationship has not been assessed earlier in the intact circulation. This difference in unloaded MV˙o 2 is substantially larger than the 11% difference that could be predicted by comparing the P:O ratios for fatty acids and glucose (35). Additional metabolic mechanisms must therefore be considered.
In the normal well-perfused and oxygenated heart, high-energy phosphate compounds are solely produced by oxidative phosphorylation, mainly through FFA oxidation, because high levels of citrate and ATP inhibit glycolysis (25, 27). Conversely, after a carbohydrate-rich meal (as during GIK administration), high levels of glucose and insulin inhibit FFA oxidation (25). Glucose decreases the rate of FFA oxidation, most likely by reducing the activity of carnitine palmitoyltransferase 1 through acetyl-CoA carboxylase catalyzed production of malonyl-CoA (19). Lactate utilization is a function of its arterial concentration at levels below 4.5 mmol/l (9). We observed a considerable uptake of lactate at baseline that tended to further increase (P = 0.08) during GIK infusion in accordance with an increment in glycolytic flux and FFA metabolism being suppressed by insulin and abundant glucose (9,25, 31). Likewise, the reduction in lactic uptake during IH infusion reflects inhibited lactate oxidation and a low rate of glycolytic flux (25). The shift in oxidative metabolisms in this study is thus a shift between the carbohydrate and the FFA pathways.
An increment in unloaded MV˙o 2is traditionally closely related to an increase in contractility (35). In our experiments, no evident difference in contractile state was observed between IH and GIK infusions. Afterload (MAP) was moderately lowered during GIK infusion, likely induced by insulin-derived vasodilatation (28). However, afterload has been shown earlier not to affect the MV˙o 2-PVA relationship (35), so neither a difference in afterload nor inotropic state is the explanation for the shift in unloaded MV˙o 2 observed between GIK and IH infusions in the present study.
Theoretically, an increased unloaded MV˙o 2 during FFA load can be caused by increased energy demand due to augmented basal metabolism or EC coupling (Fig. 1 A). In this intact animal model, these two factors were not separable because unloading and cardiac arrest is impossible during intact physiological conditions. However, reasons for an increased MV˙o 2 due to augmented basal metabolism are several. As indicated, the oxygen cost of fatty acid oxidation is higher compared with glucose, explained by stoichiometric differences (P:O ratio) (25). FFA-induced uncoupling of oxidative phosphorylation has been shown to increase oxygen consumption without a concomitant increase in ATP production (3), and a high FFA load could activate intracellular futile metabolic cycles (25). During high levels of intracellular fatty acids, the FFA-triacylglycerol cycle is reported to increase the energy demand of myocardial metabolism up to 30% (24, 32).
The main energy-consuming part of EC coupling is to reduce the cytosolic [Ca2+] by trapping into the sarcoplasmic reticulum and by extracellular extrusion (35). Burkhoff and co-workers (5) observed a 10% increase in MV˙o 2 related to EC coupling during hexanoate infusion in isolated rat hearts. This increase in MV˙o 2 could be interpreted as increased ATP usage induced by the FFA possibly due to altered Ca2+ handling. It is also reported that fatty acids increase voltage-gated Ca2+ current (14) and stimulate both the passive permeability of the sarcolemma to Ca2+ and the Na+/Ca2+ exchange (26). This would lead to an increased intracellular [Ca2+] and possibly enhanced activity of the ATP-dependent Ca2+ pumps. Taken together, an increased ATP demand of basal metabolism seems to be the most probable mechanism for a higher MV˙o 2 during FFA load, with possibly an increased rate of Ca2+ trapping as an important cofactor.
Why is it important to clarify the effect of substrates on the MV˙o 2-PVA relationship? This energetic model is used frequently to assess the effects of drugs (23,36) and disease states (13) both in experimental models (36) and in the human heart (13, 23). However, an intervention-related effect on the MV˙o 2-PVA relationship could be biased by a simultaneous change in substrate metabolism. For instance, inotropic stimulation with dobutamine increases the unloaded MV˙o 2, interpreted as increased energy consumption due to drug-induced Ca2+-increase in the myocytes (35). Alternatively, and in accordance with the present study, this effect could be induced by an increased FFA metabolism, because β-stimulation also increases the circulating levels of fatty acids (25). Likewise, mechanoenergetic inefficiency has been found in postischemic dysfunction (30) along with an increased rate of FFA oxidation (18). Thus a metabolic alteration toward a higher rate of FFA oxidation may partly explain postischemic inefficiency.
The use of GIK in the treatment of acute myocardial infarction reduces in-hospital mortality (8, 10) and has also shown promising results in postoperative left ventricular failure (12). The rationale for GIK treatment in ischemia-reperfusion is to increase the rate of ATP production from glycolysis for restoration of Ca2+homeostasis and to replenish the glycogen stores while inhibiting uptake and overload of detrimental fatty acids (7). From the present study, an “oxygen-saving” effect, without compromising contractility, should also be included in this rationale.
In the present study conductance-derived volumes were not corrected by assessing parallel volume (Vp) or by using the α-factor relating conductance to a reference volumetric method (2). The volumes reported are thus relative and not absolute. However, when SW and PVA are calculated, the maximum difference in volume per beat is used (i.e., stroke volume). Stroke volume, estimated in the present study both with transit time ultrasound and conductance volumetry, was equal during GIK and IH infusions, meaning that the α-factor for conductance correction was unchanged. Furthermore, a different Vp induced by substrate infusion will not influence the PVA, because Vp correction displaces observed volumes in a parallel manner. The difference in absolute volumes observed during GIK and IH infusions is likely explained by such a difference in Vp. This was supported by additional experiments in our laboratory where Vp was 55 ± 16 ml during GIK infusion and 64 ± 13 during IH infusion.
Unloaded MV˙o 2 obtained in vivo is an extrapolated value. However, the linearity of the MV˙o 2-PVA relationship is well established (35) also in the present study, where a high grade of linear correlation between MV˙o 2 and PVA was observed. Our estimates of unloaded MV˙o 2 should therefore be reasonably correct, although some between-subject diversity is apparent.
In conclusion, the MV˙o 2-PVA relationship is profoundly influenced by changes in myocardial substrate metabolism in vivo. In a parallel manner, the relationship was shifted upward when metabolism was shifted from glucose to FFA, giving a marked difference in unloaded MV˙o 2 as well as MV˙o 2 at comparable workloads (i.e., PVA). Our observations challenge the supposed substrate independence of the MV˙o 2-PVA relationship (35) and may be used as an argument for GIK treatment in states of oxygen deficiency. As the framework of the MV˙o 2-PVA model is used frequently in experimental and clinical studies, conclusions based on such studies may be confounded by a concomitant shift in metabolism.
The present work was supported in part by grants from the Norwegian Council on Cardiovascular Diseases and the Norwegian Research Council.
Address for reprint requests and other correspondence: C. Korvald, Dept. of Thoracic and Cardiovascular Surgery, Univ. Hospital in Tromsø, N-9038 Tromsø, Norway (E-mail:).
Professors Terje S. Larsen (Dept. of Medical Physiology) and Eivind S. P. Myhre (Depts. of Medical Physiology and Internal Medicine), University of Tromsø, are greatly acknowledged for expert commentaries.
Parts of this work previously were presented at the 71stScientific Sessions of the American Heart Association (Circulation 98: I-213, 1998).
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.
- Copyright © 2000 the American Physiological Society