Introduction

Top Abstract Introduction Materials & Methods Results Discussion References

In CHF, neurohumoral mechanisms are activated that help acutely maintain cardiac output and perfusion pressure (Packer, 1988). The continuous beta-adrenergic stimulus in CHF leads to down-regulation and functional uncoupling of cardiac beta-1 adrenoceptors, resulting in a blunted inotropic response to beta adrenoceptor agonists (Bohm et al., 1988a; Bohm et al., 1988b; Bristow et al., 1986). Similarly, the failing myocardium has a reduced response to cAMP-dependent inotropic agents such as PDE-III inhibitors (Perreault et al., 1992; Bohm et al., 1988b). Thus PDE-III inhibitors have a reduced effectiveness in improving inotropic state in the failing heart.

Agents with other mechanisms of inotropic action might have greater effect in improving LV performance in CHF. Pimobendan, a pyridazinone benzimidazole derivative, is reported to inhibit PDE-III and has the additional effect of increasing the sensitivity of the contractile proteins to activation by Ca⁺⁺ (Bohm et al., 1991; Brunkhorst et al., 1989; Hagemeijer et al., 1989; Fujino et al., 1988; Duncker et al., 1987). Although the positive inotropic effect of pimobendan has been demonstrated, the relative contributions of PDE-III inhibition and Ca⁺⁺ sensitization to its positive inotropic effect in failing myocardium are not known. Because the effect of PDE-III inhibition may be depressed in CHF, we hypothesized that the inotropic response to pimobendan due to Ca⁺⁺ sensitization may be relatively preserved after the development of CHF when compared with a more selective PDE-III inhibitor, such as amrinone (Remme, 1993). Therefore, this study was undertaken to compare the effects of pimobendan and amrinone on LV systolic and diastolic performance at rest and during exercise in conscious dogs before and after CHF.

	Materials and Methods

Top Abstract Introduction Materials & Methods Results Discussion References

Instrumentation

Seven healthy, adult, heart worm-negative mongrel dogs (weight 25-36 kg) were instrumented under anesthesia after induction with xylazine (2 mg/kg i.m.) and sodium thiopental (6 mg/kg i.v.) and maintained with halothane (0.5-2.0%). They were intubated and ventilated with oxygen-enriched room air to maintain arterial oxygen pressure greater than 100 mm Hg and pH between 7.38 and 7.42. A sterile left lateral thoracotomy was performed, and pericardium was widely opened. Micromanometer pressure transducers (Konisberg Instruments, Inc., Pasadena, CA) and polyvinyl catheters (1.1 mm I.D.) for transducer calibration were inserted into the LV through an apical stab wound and into the LA. Three pairs of ultrasonic crystals (5 MHz) were implanted in the endocardium of the LV to measure the anterior-to-posterior, septal-to-lateral, and base-to-apex (long-axis) dimensions, using the method previously described from our laboratory (Cheng et al., 1993). Hydraulic occluder cuffs were placed around the inferior and superior venae cavae. A pacing lead was attached to the right ventricle and connected to a programmable pacemaker (model 8329, Medtronic Inc., Minneapolis, MN) implanted s.c. All wires and tubings were exteriorized through the posterior neck.

Data Collection

Studies were performed after full recovery from instrumentation (from 10 days to 2 weeks after surgery) with the dogs standing and then running on a motorized treadmill (model 1849C, Quinton Inc., Seattle, WA). The LV and LA catheters were connected to pressure transducers (Statham P23Db, Gould, Cleveland, OH) calibrated with a mercury manometer. The signal from the micromanometers was adjusted to match that of the catheters. The LA micromanometer was adjusted to match LV pressure at the end of long periods of diastasis.

The analog signals were recorded on an eight-channel chart-recorder (Astro-Med, West Warwick, RI), digitized with an on-line analog-to-digital converter (Data Translation Devices, Marlboro, MA) at 200 Hz and were stored on a magneto-optical disk memory system.

Experimental Protocol

The effects of amrinone and pimobendan were assessed in seven dogs before and after CHF. The order in which the drugs were studied was randomly determined, and at least 24 hr elapsed between studies.

Studies before CHF. Steady-state data and data during transient caval occlusion were recorded at rest while the animals stood on a motorized treadmill. Three sets of variably loaded P-V loops were generated by caval occlusion. The animals then ran on the treadmill. The treadmill speed was gradually increased over 1 to 2 min from 2.5 miles/hr to the maximum level tolerated for exercise (5.5-8.0 miles/hr). The animals exercised at this level until they could no longer keep up with the treadmill. The total exercise time ranged from 7 to 12 min. We analyzed the data recorded at rest and during the last minute of exercise. After a 30 min rest after control exercise, amrinone (1 mg/kg i.v. followed by 10 µg/kg/min infusion) or pimobendan (0.25 mg/kg i.v.) was administered. Forty minutes after drug administration, steady-state and caval occlusion data at rest were collected, and then the treadmill exercise protocol was repeated and steady-state data were collected.

Studies during the development of CHF. After completion of the base-line exercise studies, the pacemaker rate was adjusted, using the external magnetic control unit, to 220 to 250 beats per minute. Three times per week, the pacemaker rate was adjusted below the spontaneous rate. The animal was allowed to equilibrate for 30 minutes, and then data were collected. After each study, pacing rate was returned to 220 to 250 beats per minute. After pacing for 4 to 5 weeks, when the LV P_ED during the nonpaced period had increased by more than 15 mm Hg over the prepacing control level, CHF data were obtained. This level of CHF was chosen because the animals had begun to show clinical evidence of CHF (anorexia, mild ascites, and pulmonary congestion) but were still able to exercise.

Studies after the onset of CHF. Studies in dogs with CHF were performed after the animal stabilized for at least 30 min after discontinuing pacing. Steady-state and caval occlusion data were collected with the animal standing at rest. Then the animal ran on the treadmill as the speed was increased and adjusted to the maximum tolerated steady-state level, and data were collected while the animal was running. After CHF, the maximum level of exercise was decreased to 3.5 to 6.0 miles/hr. The total exercise duration was also reduced (range from 4 to 7 min). Then, after a 30 min rest, the same amount of amrinone (1 mg/kg i.v. followed by 10 µg/kg/min infusion) or pimobendan (0.25 mg/kg i.v.), as used in the studies conducted before CHF, was administrated. Forty minutes after drug administration, resting steady-state and caval occlusion data were collected, and then treadmill exercise was performed.

Data Processing and Analysis

V_LV was calculated as a modified general ellipsoid using the following equation: where D_AP is the anterior-to-posterior LV diameter, D_SL is the septal-to-lateral LV diameter, and D_LA is the long-axis LV diameter. We have previously demonstrated that this method gives a consistent measure of V_LV despite changes in LV loading conditions, configurations and HR (Cheng et al., 1993; Little et al., 1988; Little et al., 1989). To account for respiratory changes in intrathoracic pressure, steady-state measurements were averaged over the 12- to 15-sec recording period that spanned multiple respiratory cycles. End-diastole was defined as the relative minimum of LV pressure occurring after the A wave. End-systole was defined as the upper left corner of the LV P-V loop. The time of mitral valve opening was defined to be when LV pressure fell below LA pressure. LV end-diastolic, end-systolic and minimum pressure were measured. We also measured LV V_ED and V_ES. The mean LA pressure was determined.

The derivatives of LV pressure (dP/dt) and LV volume (dV/dt) were calculated using the five-point Lagrangian method. Stroke volume was calculated as V_ED minus V_ES. Cardiac output was determined as stroke volume times HR. We also calculated SW by point-by-point integration of the LV P-V loop for each beat. The rate of LV relaxation was analyzed by determining the time constant of the isovolumic fall of LV pressure. LV pressure from the time of minimum dP/dt until mitral valve opening was fit to an exponential equation: where P is LV pressure, t is time and P_A, P_B and T are constants determined by data. Although the fall in isovolumic pressure is not exactly exponential, the time constant, derived from the exponential approximation, provides an index of the rate of LV relaxation (Gilbert and Glantz, 1989). The effective E_A was calculated as LV P_ES divided by stroke volume. We also calculated TSR as P_ES divided by cardiac output. Ejection fraction was calculated as stroke volume divided by V_ED.

Analyses of LV P-V loop during caval occlusion. Only caval occlusions that produced a fall in LV P_ES of approximately 30 mm Hg were analyzed. Premature beats and the subsequent beat were excluded from analysis.

Post-mortem Evaluation

At the conclusion of the studies, the animals were sacrificed by lethal injections of sodium thiopental (100 mg/kg i.v.), and the heart was examined to confirm the proper position of the instrumentation.

Statistical Analysis

Statistical comparisons were made with Student's t test for paired observations and analysis of variance with the Bonferroni method of multiple-paired comparisons as appropriate. Significance was accepted when P < .05. Data for steady state are expressed as mean ± S.D.; values for LV P-V relations are expressed as mean ± S.E.M.

	Results

Top Abstract Introduction Materials & Methods Results Discussion References

Effects of Pimobendan and Amrinone Administered Before CHF at Rest and During Exercise

Steady-state measurements. Steady-state hemodynamic changes produced with amrinone and pimobendan before CHF at rest and during exercise are summarized in table 1. At rest, both amrinone and pimobendan (fig. 1) evoked significant (P < .05) and equivalent decreases in LV P_ES and TSR. Amrinone and pimobendan also produced similar significant (P < .05) increases in HR (121 ± 11 vs. 132 ± 11, 122 ± 9 vs. 131 ± 9 bpm), dP/dt_max and ejection fraction but caused no significant changes in stroke volume. During exercise, both drugs produced similar increases in HR (188 ± 16 vs. 194 ± 15, 182 ± 14 vs. 193 ± 17 bpm) and similar and significant decreases in minimum LV pressure. Compared with exercise without drugs, both amrinone and pimobendan did not significantly alter the SV or cardiac output. As shown in table 1, at rest both amrinone and pimobendan caused a significant decrease in T (27.9 ± 2.8 vs. 24.3 ± 3.5; 27.1 ± 4.8 vs. 21.1 ± 3.6 msec) and had no effect on the peak rate of LV filling (dV/dt_max).

View larger version (K): [in this window] [in a new window]   Fig. 1.   Steady-state LV P-V loops obtained at rest (upper panels) and during exercise (lower panels) in a normal dog before (control) and after administration of amrinone or pimobendan. Both agents shifted the P-V loops toward the left at rest and during exercise.

Pressure-volume analysis. As shown in table 2 and fig. 2, in normal dogs, amrinone and pimobendan produced similar significant increases in the slopes of the P_ES-V_ES relation (3.4 ± 1.5 vs. 4.2 ± 1.1 mm Hg/ml), the dP/dt_max-V_ED relation (37.7 ± 30.9 vs. 41.8 ± 18.4 mm Hg/sec/ml) and the SW-V_ED relation (20.5 ± 11.6 vs. 26.3 ± 4.4 mm Hg). There were also significant leftward shifts of all three relations in the physiologic range, manifested by significant decreases in V_100,ES (24.0 ± 8.2 vs. 21.5 ± 7.8; 24.6 ± 6.8 vs. 20.1 ± 6.7 ml), V_2000,dP/dt and V_2000,SW (46.3 ± 7.1 vs. 42.0 ± 7.0; 45.4 ± 5.6 vs. 37.4 ± 6.7 ml). The increases in slopes and leftward shifts of LV P-V relations with amrinone and pimobendan indicate that they produced a similar enhancement of LV contractile performance in normal dogs.

View larger version (K): [in this window] [in a new window]   Fig. 2.   LV P-V loops produced by transient caval occlusion during control and after administration of amrinone (left panel) or pimobendan (right panel) in a normal dog. The LV end-systolic P-V relations are indicated by lines. Both amrinone and pimobendan shifted the LV end-systolic P-V relations toward the left with an increase in slope, indicating similar positive inotropic action in this normal animal.

Effect of pacing-induced CHF. After the development of CHF at rest, the mean P_ED increased from 11.3 ± 2.9 to 27.7 ± 4.5 mm Hg (P < .05) (tables 1 and 3). The minimum LV pressure (0.6 ± 2.3 vs. 5.6 ± 3.3 mm Hg, P < .05) and mean LA pressure (6.9 ± 1.7 vs. 18.6 ± 3.8 mm Hg, P < .05) also increased. The LV V_ES and V_ED increased, whereas cardiac output was decreased because of a fall in stroke volume. The time constant T increased (27.5 ± 3.8 vs. 35.3 ± 3.0 msec, P < .05). LV contractility was also significantly impaired, as indicated by the decreased slopes and rightward shifts of the P-V relations (table 4). Furthermore, after CHF, the LV response to exercise was altered. As shown in figure 3, with CHF, exercise caused significant elevation of minimum LV pressure and P_ED. The early diastole portion of the LV P-V loop was shifted upward. T significantly increased.

View larger version (K): [in this window] [in a new window]   Fig. 3.   LV P-V loops obtained at rest (upper panels) and during exercise (lower panels) in the same animal after development of pacing-induced CHF. Amrinone had little effect at rest or during exercise. In contrast, pimobendan produced a leftward shift of the LV P-V loop and reduced early diastolic LV pressure.

Effects of Pimobendan and Amrinone in Dogs with CHF at Rest and During Exercise

Steady-state data measurements. The steady-state hemodynamic response produced by amrinone and pimobendan at rest and during exercise after CHF is summarized in table 3. After CHF, both amrinone and pimobendan produced a similar decrease in E_A and TSR. After CHF, pimobendan markedly increased SV. Therefore, P_ES did not decrease with pimobendan after CHF. There were no significant differences in the HR. However, only pimobendan caused a marked improvement in LV diastolic performance as indicated by a significant decrease in T (1.0 ± 0.7 vs. 9.5 ± 4.1 msec, P < .05), an increase in dV/dt_max and a decrease in minimum LV pressure. Amrinone did not improve these parameters. Furthermore, pimobendan produced a greater reduction in P_ED and mean LA pressure. These effects of pimobendan persisted during exercise (table 3).

Pressure-volume analysis. The changes of P-V relations by amrinone and pimobendan after CHF are summarized in table 4. Typical examples of the effect of amrinone and pimobendan on variably loaded P-V relations from one animal with CHF are shown in figure 4. After CHF, amrinone caused a slight leftward shift of the LV P_ES-V_ES relation, but the slope of this relation was relatively unchanged. Amrinone produced no significant change in the slope of the dP/dt_max-V_ED relation (35.5 ± 6.7 vs. 42.2 ± 9.1 mm Hg/sec/ml) or of the SW-V_ED (52.2 ± 6.3 vs. 61.4 ± 11.3 mm Hg) relation. In contrast, pimobendan produced a markedly leftward shift of the P_ES-V_ES relation with an increased slope (3.7 ± 0.3 vs. 3.8 ± 0.4; 3.9 ± 0.6 vs. 5.7 ± 0.4 mm Hg/ml, P < .05). In addition, pimobendan increased the slope of the dP/dt_max-V_ED relation (38.8 ± 8.4 vs. 57.1 ± 6.6 mm Hg/sec/ml, P < .05) and that of the SW-V_ED relation (53.6 ± 4.5 vs. 83.6 ± 10.4 mm Hg, P < .05).

View larger version (K): [in this window] [in a new window]   Fig. 4.   LV P-V loops produced by bicaval occlusion after the development of pacing-induced CHF in the same animal. Amrinone produced little change in the LV end-systolic P-V relation (shown by line), whereas the slope of the relation was increased by pimobendan. This indicates that pimobendan's inotropic effect persisted after CHF, whereas the effect of amrinone was diminished.

Left Ventricular-Arterial Coupling and Work Efficiency of the LV

We evaluated left ventricular-arterial coupling and the SW/PVA ratio in normal dogs and dogs with CHF at rest. Data are summarized in table 5, A and B. SW was altered by neither amrinone nor pimobendan in normal dogs. Amrinone and pimobendan significantly increased the E_ES/E_A ratio to a similar extent (0.70 ± 0.24 vs. 0.79 ± 0.22, P = NS, amrinone vs. pimobendan) in normal animals. The SW/PVA ratio was also increased by both drugs to similar magnitude (0.12 ± 0.03 vs. 0.14 ± 0.06, P = NS).

In dogs with CHF, pimobendan significantly increased SW; however, amrinone did not. The E_ES/E_A ratio was significantly increased by both drugs (0.10 ± 0.05 vs. 0.44 ± 0.08, P < .05). The SW/PVA ratio was also significantly augmented by both drugs, though the magnitude was higher with pimobendan than with amrinone (0.05 ± 0.03 vs. 0.18 ± 0.05, P < .05).

	Discussion

Top Abstract Introduction Materials & Methods Results Discussion References

We studied the acute effect of i.v. doses of pimobendan and amrinone that produced equivalent inotropic and arterial vasodilatory actions in normal, instrumented animals at rest and during exercise. After the induction of pacing-induced CHF, the vasodilatory effects of amrinone and pimobendan persisted. However, after CHF, amrinone's inotropic effects at rest and during exercise were markedly attenuated. In contrast, pimobendan's inotropic effect persisted after CHF. Thus pimobendan, a Ca⁺⁺ sensitizer with some PDE-III inhibitor effects, was more effective than amrinone, a relatively pure PDE-III-inhibitor, in enhancing LV contractile state in CHF.

The bipyridine derivative amrinone produces positive inotropic effects and arterial vasodilation under normal circumstances (Honerjager, 1991). These effects are mediated by inhibition of PDE-III in cardiomyocytes (Honerjager, 1991; Morgan et al., 1986) and in vascular smooth muscle cells (Morgan et al., 1986; Honerjager et al., 1981). The increased cAMP in cardiomyocytes induced by PDE-III inhibition enhances the slow Ca⁺⁺ inward current, producing a larger Ca⁺⁺ transient (Honerjager, 1991). This increased Ca⁺⁺ transient results in increased contractile force. Phosphorylation of phospholamban by cAMP-dependent phosphokinase also enhances Ca⁺⁺ uptake into the sarcoplasmic reticulum (Honerjager, 1991), speeding the rate of relaxation. Vascular smooth muscle cells contain a cAMP-dependent protein kinase, which activates a sarcolemmal Ca⁺⁺ pump (Honerjager, 1991; Morgan et al., 1986). Thus increased cAMP in vascular smooth muscle cells decreases intracellular Ca⁺⁺, resulting in vasodilation (Honerjager, 1991; Morgan et al., 1986). Amrinone does not have a Ca⁺⁺-sensitizing effect in cardiomyocytes (Berger et al., 1985).

The benzimidazole derivative pimobendan is also an inotropic agent with vasodilating properties (Bohm et al., 1991; Brunkhorst et al., 1989; Hagemeijer et al., 1989; Fujino et al., 1988; Duncker et al., 1987). Pimobendan decreases PDE-III activity in cardiomyocytes and vascular smooth cell, producing pimobendan's vasodilatory action (Fujimoto, 1994; Fujimoto and Matsuda, 1990). The positive inotropic action of this drug is at least partly associated with potentiation of the slow Ca⁺⁺ inward current in cardiomyocytes (Morgan et al., 1986; Hagemeijer et al., 1989). However, Berger et al. (1985) reported that pimobendan inhibits PDE-III activity only by 20 to 30% at the concentration producing a maximal positive inotropic effect in guinea pig papillary muscles, a result that suggests an additional mechanism for its inotropic action. Several investigators have shown that pimobendan increases the Ca⁺⁺ sensitivity of cardiac myofilament by a direct effect on Ca⁺⁺-binding affinity of myofilament troponin C (Fujino et al., 1988; van Meel, 1987; Duncker et al., 1987; Ruegg et al., 1984). Pimobendan is demethylated rapidly in the body to an active metabolite, UD-CG 212 C1 (Honerjager, 1991). UD-CD 212 C1 elicits positive inotropic and vasodilating effects mediated by inhibition of PDE-III (Endoh et al., 1991; Hagemeijer et al., 1989; Duncker et al., 1987), and Ca⁺⁺ sensitization under some circumstances (van Meel et al., 1995a; van Meel et al., 1995b).

In failing myocardium, the positive inotropic response to PDE-III inhibitors is impaired (Perreault et al., 1992; Bohm et al., 1988b; Feldman et al., 1987). The reduced response to those agents may be due to reduced basal cAMP formation in the failing heart (Perreault et al., 1992; Bohm et al., 1988b), beta-adrenergic receptor down-regulation (Bohm et al., 1988a; Bristow et al., 1986) and an increase in the Gi, -subunit of the guanine nucleotide-binding protein that inhibits adenyl cyclase activity in myocardial membranes (Marzo et al., 1991; Bohm et al., 1990; Feldman et al., 1988).

In a result consistent with these observations, we found that the inotropic effects of amrinone were decreased after CHF. However, pimobendan's positive inotropic effect persisted. Thus it appears unlikely that PDE-III inhibition from pimobendan and its metabolite contributed to pimobendan's inotropic effect after CHF. Instead, pimobendan's inotropic effect after CHF appears to be due to Ca⁺⁺ sensitization.

Pimobendan and amrinone produced equivalent amounts of arterial vasodilation at rest and during exercise both before and after CHF. This suggests that the vasodilator effects in both drugs, which are mediated by PDE-III inhibition in vascular smooth muscle, are not attenuated in CHF. This is consistent with previous reports that amrinone and pimobendan decrease arterial resistance even in severe CHF (Katz et al., 1992; Baumann et al., 1989; Konstam et al., 1986; Bayliss et al., 1983).

During exercise (both normally and after CHF), the heart is exposed to increased adrenergic stimulation (Chidsey et al., 1962). The increased adrenergic stimulation may enhance the inotropic response to PDE-III inhibitors (Cheng et al., 1992; Perreault et al., 1992). However, we observed that after CHF, pimobendan had a much greater inotropic effect during exercise than did amrinone.

Before CHF, amrinone and pimobendan had similar effects on parameters of LV diastolic performance (minimum LV pressure, T and dV/dt_max) both at rest and during exercise. After CHF, pimobendan produced a greater reduction than amrinone in minimum LV and LA pressure at rest and during exercise. Both amrinone and pimobendan increased the rate of relaxation after CHF at rest and during exercise. This may have been at least partially due to a reduction in afterload. However, pimobendan's effects were greater than those of amrinone.

Our results should be compared with those of Asanoi et al. (1994), who also studied the effect of pimobendan in dogs with pacing-induced CHF. In agreement with our findings, they found that pimobendan's inotropic and lusitropic effects persisted after CHF. In contrast, they found that the inotropic effects were reduced compared with control.

In conclusion, our study suggests that the inotropic response to PDE-III inhibitors is attenuated in CHF, whereas the response to Ca⁺⁺ sensitizer is preserved. Thus the drugs that increase contractile protein Ca⁺⁺ sensitivity may be more effective in producing acute inotropic support to the failing heart.