Left Ventricular Contractility Predicts How the End-Diastolic Pressure-Volume Relation Shifts During Pacing-Induced Ischemia in Dogs
Background Two types of ischemia, pacing-induced and coronary occlusion–induced, have different effects on left ventricular diastolic properties. During pacing-induced ischemia, the diastolic pressure-volume relation is said to shift upward, whereas during coronary occlusion, it is said to shift rightward or downward. However, recent studies have shown that the relation can shift in any direction during both types of ischemia. The purpose of this study was to identify determinants of the shift of the end-diastolic pressure-volume relation (EDPVR) during pacing-induced ischemia.
Methods and Results We retrospectively analyzed 46 pacing-induced ischemia experiments performed in 15 open-pericardium anesthetized dogs. Pacing ischemia was induced by constricting left anterior descending and left circumflex coronary arteries and pacing the left atrium at 150 to 180 beats per minute for 3 minutes. Left ventricular volume was measured with a conductance catheter. Hemodynamics were recorded during baseline, coronary stenosis, rapid pacing, and pacing-induced ischemia (immediately after rapid pacing). For each condition, hemodynamics were recorded in steady state and then during a brief inferior vena caval occlusion (except for during rapid pacing) to obtain left ventricular end-diastolic and end-systolic pressure-volume relations. The shift of the EDPVR from coronary stenosis to pacing-induced ischemia was assessed by an upward shift index (end-diastolic pressure during pacing-induced ischemia minus the pressure during coronary stenosis at the largest end-diastolic volume common to both conditions, SI-S) and a rightward shift index (the largest end-diastolic volume during pacing-induced ischemia minus the largest volume during coronary stenosis, ΔEDVI-S). The index of left ventricular contractility, the end-systolic elastance (Ees), or the slope of the dP/dtmax–end-diastolic volume relation (dE/dtmax) during pacing-induced ischemia was the strongest determinant of the magnitude of SI-S and ΔEDVI-S and thus of the shift of the EDPVR. As Ees or dE/dtmax decreased, SI-S decreased and ΔEDVI-S increased.
Conclusions Our results suggest that left ventricular contractility is the best determinant of the shift of the EDPVR during pacing-induced ischemia. The more left ventricular contractility decreases, the more the EDPVR shifts downward and rightward.
The left ventricular diastolic pressure-volume relation changes when the heart is made ischemic. The particular response of the left ventricular diastolic pressure-volume relation has been thought to depend on how ischemia is induced. When ischemia is induced by rapid pacing in the presence of critical coronary stenosis (demand ischemia), the diastolic pressure-volume (or pressure-length) relation shifts upward.1 2 3 4 5 When ischemia is induced by coronary occlusion (supply ischemia), the relation shifts rightward or downward.4 5 6 7 Recent studies, however, have shown that a rightward shift can also occur during pacing-induced ischemia,8 and an upward shift can occur during coronary occlusion.9 10 Similarly, in earlier experiments, we found shifts in various directions (upward, rightward, and downward) and of various magnitudes during pacing-induced ischemia.11 Determinants of the different directions and magnitudes of the shift of the diastolic pressure-volume relation during the same type of ischemia are still unclear. To identify determinants of the shift of the diastolic pressure-volume relation during pacing-induced ischemia, we retrospectively analyzed data from our earlier experiments.11
Although most studies have assessed the shift of the diastolic pressure-volume relation by comparing pressure-volume or pressure-length loops of steady-state beats, the magnitude of the shift cannot be quantitatively assessed by comparing these loops when left ventricular volumes or lengths are not identical. Therefore, we assessed the shift of the end-diastolic pressure-volume relation (EDPVR) obtained during a brief vena caval occlusion to get identical volumes and examined the relation between the shift of the EDPVR and various hemodynamic variables.
For this study, we analyzed data from an experiment that was designed to examine whether gadolinium (an inhibitor of stretch-activated ion channels) attenuates the upward shift of the EDPVR during pacing-induced ischemia in dogs.11 The experiment was approved by the Committee on Animal Research at the University of California and conforms to the guiding principles of the American Physiological Society.
Fifteen mongrel dogs of either sex (weight, 26.7±4.0 kg, mean±SD) were anesthetized with 100 mg/kg IV α-chloralose, intubated, and ventilated with a mixture of air and oxygen. Anesthesia was supplemented by 10 mg IV morphine every hour. Arterial blood gases were analyzed periodically and the respiratory rate and oxygen flow adjusted to keep pH at 7.35 to 7.45, Po2 >100 mm Hg, and Pco2 at 35 to 50 mm Hg. A left thoracotomy was performed at the sixth intercostal space, and the pericardium was opened wide. The left anterior descending coronary artery (LAD) just distal to the bifurcation of the first diagonal branch and the left circumflex coronary artery (LCx) at the proximal segment were dissected, and transit-time ultrasonic flow probes (Transonic Systems Inc) and adjustable vessel occluders (similar to small C-clamps)12 were placed around these arteries. To measure left ventricular segment lengths, two pairs of miniature flat-plate ultrasonic crystals (Triton Technology) were implanted in the left ventricular wall close to the endocardium in a circumferential plane, one pair in the anterior wall, perfused by the LAD, and the other pair in the lateral wall, perfused by the LCx. High-fidelity 5F pressure micromanometer catheters (Millar Instruments) were inserted into the left ventricle, right ventricle, and left atrium via the femoral artery, jugular vein, and pulmonary vein, respectively. A pair of pacing electrodes was attached to the left atrial appendage. An 8F Fogarty venous thrombectomy catheter was inserted into the inferior vena cava from the femoral vein to provide a transient vena caval occlusion. A 7F eight-electrode conductance catheter (Webster Laboratories) was inserted into the left ventricle through the left ventricular apex to measure left ventricular volume. The conductance catheter was connected to a signal processor (Stiching Leycom Sigma-5). All signals (pressures, volume, coronary flows, segment lengths, and ECG) were continuously monitored on a multichannel oscillograph throughout the experiment.
After the dogs were instrumented, propranolol (0.5 mg/kg) was administered by slow bolus injection to prevent ventricular fibrillation during rapid pacing,13 as was routinely used in previous studies of pacing-induced ischemia.1 2 3 4 5 8 This small dose had little effect on heart rate or hemodynamics. Then, in 14 dogs, the left atrium was paced at 100 beats per minute (bpm) after 1 mg/kg UL-FS4914 was injected to reduce the heart rate. In the 15th dog (dog 443), for which UL-FS49 was not available, the baseline heart rate was 133 bpm. After each dog’s hemodynamics stabilized, data for baseline were recorded.
As in most of the previous studies,1 2 3 4 5 8 12 pacing ischemia was induced by 3 minutes of rapid pacing in the presence of coronary stenosis in both the LAD and the LCx. To create coronary stenosis, the vessel occluders were tightened on both coronary arteries until impaired systolic shortening of the segment lengths was observed, then the occluders were slightly released so that the stenosis had no detectable effect on systolic shortening.5 8 12 Data for coronary stenosis were then recorded. Most of the hemodynamic data did not change from baseline to coronary stenosis (see “Results”), indicating that coronary stenosis alone did not induce ischemia. Next, to induce ischemia, the left atrium was paced at 150 to 180 bpm for 3 minutes. When AV block occurred at high pacing rate (<10% of the time), we paced the left ventricle. The pacing rate was then abruptly returned to 100 bpm (pacing was stopped in dog 443), and 2 or 3 seconds later, data for pacing-induced ischemia were recorded.
Because the original study was designed to examine whether gadolinium attenuates the upward shift of the EDPVR during pacing-induced ischemia,11 in that study we first examined whether the relation shifted upward during pacing-induced ischemia before the gadolinium injection. For that examination, we used data-processing software (described in “Data Acquisition”) that compared the diastolic pressure-volume loops during pacing-induced ischemia with those during coronary stenosis. When there was an upward shift large enough to examine whether gadolinium would attenuate the upward shift, we went on to the next step, an injection of gadolinium. When a large enough upward shift did not occur, we waited at least 10 minutes after the previous rapid pacing for hemodynamics to stabilize, then performed another pacing-induced ischemia experiment, in which we changed the extent of coronary stenosis or the rate of rapid pacing. To change the extent of coronary stenosis, we tightened the occluders slightly if the diastolic pressure-volume loop did not shift either upward or rightward and released the occluders slightly if the loop shifted rightward or downward. We changed the rate of rapid pacing between 150 and 180 bpm when changing the extent of coronary stenosis did not produce an upward shift; we increased the rate if the pressure-volume loop did not shift either upward or rightward and decreased the rate if the EDPVR shifted rightward or downward. We repeated this procedure until a large upward shift of the EDPVR was observed or was not possible. A large upward shift was observed in 5 dogs: on the first pacing-induced ischemia attempt in 2 dogs, on the second attempt in 1 dog, on the third attempt in 1 dog, and on the fifth attempt in 1 dog. In the other 10 dogs, we did not observe a large upward shift and gave up after 1 to 9 pacing-induced ischemia experiments because of ventricular fibrillation. We did not use data after ventricular fibrillation (when it occurred) because ventricular fibrillation may alter the myocardial oxygen supply-demand balance and thereby influence left ventricular diastolic properties.
After the experiment, the dog was given an injection of pentobarbital intravenously followed by a lethal injection of concentrated potassium chloride to the left ventricle. The heart was then removed, and the left ventricle was weighed. Left ventricular weight was 118±20 g (mean±SD) (range, 74 to 140 g).
For the present study, data from 46 pacing-induced ischemia experiments (before the gadolinium injection) performed in the 15 dogs were used. The number of experiments in each dog ranged from 1 to 9 (3.1±2.1). Each experiment consisted of baseline, coronary stenosis, rapid pacing, and pacing-induced ischemia. Because there was only one baseline in each dog, the same baseline was used for comparison with all pacing-induced ischemia experiments in a given dog.
During baseline, coronary stenosis, and pacing-induced ischemia, data were recorded for 25 seconds after the ventilator was stopped at end expiration. The first 7 seconds of data were recorded in steady state to obtain steady-state hemodynamic variables and the last 18 seconds during vena caval occlusion to obtain EDPVRs and end-systolic pressure-volume relations (ESPVRs). Earlier experiments on 6 dogs confirmed that the left ventricular pressure-volume relation is stable for at least 30 seconds after rapid pacing.11 This stability allowed us to determine left ventricular EDPVRs and ESPVRs over a wide range of volumes with vena caval occlusion during pacing-induced ischemia. During rapid pacing, data were recorded in steady state for 5 seconds after 2 minutes of pacing.
The data recorded were left ventricular pressure, right ventricular pressure, left atrial pressure, coronary blood flows (in LAD and LCx), left ventricular volume, and left ventricular segment lengths (anterior and lateral). These data were digitized on-line by a personal computer (Macintosh II) with a 12-bit analog-to-digital converter (National Instruments NB-MIO-16) using data-processing software (labview 2, National Instruments) at a 200-Hz sampling rate. The data were then uploaded to a MIPS minicomputer and analyzed with software developed in our laboratory. Premature ventricular contractions and the subsequent beat were excluded from the analysis.
Left Ventricular Volume
To avoid introducing errors due to the uncertainties in existing methods that estimate the parallel conductance volume and slope, α, that relates conductance catheter volume to true left ventricular volume in individual dogs,15 we used a standard calibration function for the conductance catheter derived from a similar preparation in our laboratory16 :
where V is the corrected left ventricular volume and VCC is the uncorrected conductance catheter volume (both in milliliters). Because this calibration was based on volume computed with epicardial crystals,16 it includes (constant) left ventricular wall volume. As a result, the reported volumes are larger than those of the left ventricular cavity alone, and the ejection fractions are smaller. Values of stroke volume, stroke work, and ESPVR and EDPVR in this study would be the same as those obtained from the cavity volume, because the epicardial volume changes in parallel with the cavity volume. Use of this calibration in the present study is based on the assumption that the parallel conductance does not change during pacing-induced ischemia. Although we could not measure parallel conductance during pacing-induced ischemia because pacing-induced ischemia was stable only 30 seconds, which did not allow us to record data and measure parallel conductance volume during the same pacing-induced ischemia, Kass et al17 and Applegate18 reported that parallel conductance volume did not change significantly during coronary occlusion, which produces more severe changes in hemodynamics than pacing-induced ischemia.
Variables in Steady-State Beats
Hemodynamic variables in steady-state beats were calculated as the mean of the variables for the first seven beats in each recording.
Stroke volume was calculated as
where SV is stroke volume, Ved is end-diastolic volume, and Ves is end-systolic volume. End diastole was defined as the starting point of the rapid upstroke of the first derivative of left ventricular pressure (dP/dt). End-systole was defined as the point of maximum systolic elastance, computed by the iteration method of Kono et al.19
Stroke work was calculated as:
where SW is stroke work, P is left ventricular pressure, and V is left ventricular volume, and the integral is from end diastole to the end diastole of the next beat.
To account for differences in heart size among dogs, coronary flow (LAD plus LCx mean flow), left ventricular volume, stroke volume, and stroke work were divided by left ventricular weight, and the values per 100 g of left ventricular weight were used for analysis.
Percent systolic shortening of the segment length was calculated as
where SS is the percent systolic shortening, EDL is end-diastolic segment length, and ESL is end-systolic segment length. Satisfactory segmental length data measured by ultrasonic crystals were available in 38 of 46 experiments.
The time constant of left ventricular isovolumic relaxation was obtained by the linear regression
where P is left ventricular pressure, τ is the time constant of left ventricular isovolumic relaxation, t is time from the minimum dP/dt (dP/dtmin) point, and P0 is the estimated left ventricular pressure at the time of dP/dtmin.20 Left ventricular pressure data from the time of dP/dtmin to a time at which left ventricular pressure equaled end-diastolic pressure plus 5 mm Hg were used in this calculation. Although it is well established that the curve of ln P versus t is slightly nonlinear (concave down),21 the exponential time constant estimated from the pressure fall during isovolumic relaxation is a good way to quantify how rapidly the pressure falls.21 22
Left Ventricular Contractility
Left ventricular contractility was assessed by means of two load-independent indices, the slope of ESPVR and the slope of the dP/dtmax–Ved relation.
To define the ESPVR so that we could calculate end-systolic elastance (Ees) as an index of left ventricular contractility, we applied linear and curvilinear regression to the left ventricular end-systolic points of the pressure-volume loops (Fig 1⇓, left) during vena caval occlusion. The linear ESPVR was defined as
where Pes is the end-systolic pressure, Ves is the end-systolic volume, and Vd is the volume axis intercept.23 Ees and Vd were estimated by use of the iteration procedure described by Kono et al.19 The curvilinear ESPVR was defined by the quadratic equation24
The coefficients a, b, and c were estimated with polynomial regression. The coefficient a represents a measure of the curvilinearity of the relation. Although it was proposed to use the slope of the curvilinear relation at Pes=0 mm Hg as the index of contractility,24 we did not have data at low enough pressures to estimate precisely the slope at 0 mm Hg in all ESPVRs, especially during pacing-induced ischemia. To avoid problems with extrapolation of the curvilinear ESPVR to zero pressure, we calculated local elastance (dPes/dVes or E′es) located on the ESPVR at a common pressure within the measured range.25 As the common pressure, we used the lowest end-systolic pressure that was commonly observed in all ESPVRs in a given dog (Fig 2⇓, left). E′es, the slope of the curvilinear ESPVR at the end-systolic pressure of Pm, is
The quadratic equation provided a significantly better fit of the ESPVR points than did linear regression (the coefficient a in the quadratic equation was significantly different from 0) in 85 of 107 ESPVRs. Because it is now accepted that the ESPVR is curvilinear,24 26 27 28 we used E′es obtained from the curvilinear regression in these 85 ESPVRs. In 22 other ESPVRs for which the quadratic term was not significant, we used Ees from the linear regression.
The dP/dtmax–Ved relation was defined as
where dE/dtmax is the slope of the dP/dtmax–Ved relation, Ved is end-diastolic volume, and VdP/dt is the intercept with the volume axis.29 The dP/dtmax–Ved data during the vena caval occlusion were fit by linear regression (Fig 2⇑, right).
We expressed Ees and dE/dtmax during coronary stenosis and during pacing-induced ischemia as percentages of the values during baseline.25
End-Diastolic Pressure-Volume Relation
To define the EDPVR, we applied the exponential equation to left ventricular end-diastolic points of the pressure-volume points during vena caval occlusion:
where Ped is left ventricular end-diastolic pressure and V is left ventricular end-diastolic volume30 (Fig 1⇑, right). The parameters P0, α, and β were estimated with the Marquard-Levenberg algorithm. This equation was used only to describe the EDPVR curve; the parameters P0, α, and β do not have any direct physiological significance.31
Shift of the EDPVR
There were four qualitative responses of the EDPVR between coronary stenosis and pacing-induced ischemia. The first response was that the EDPVR did not change discernibly from coronary stenosis to pacing-induced ischemia (Fig 3⇓, minimal shift). The second response was that the EDPVR during pacing-induced ischemia moved upward compared with the EDPVR during coronary stenosis (Fig 3⇓, upward shift). The third response was that the EDPVR during pacing-induced ischemia moved both rightward and upward along the EDPVR curve during coronary stenosis and thus that the EDPVR assessed at the common volume was not shifted either upward or downward (Fig 3⇓, rightward shift). The fourth response was that the EDPVR during pacing-induced ischemia moved downward (and often rightward) compared with the EDPVR during coronary stenosis (Fig 3⇓, downward shift).
It is worth noting that assessing the shift of the diastolic pressure-volume relation quantitatively by comparing pressure-volume loops in steady-state beats was difficult because left ventricular volume increased in most cases (Fig 3⇑, left panels). For example, diastolic pressure-volume loops associated with rightward shift and downward shift of the EDPVRs in Fig 3⇑ appear to shift upward because the early diastolic portion of the loops during pacing-induced ischemia was located higher than the loops during coronary stenosis in an overlapping volume range. For these reasons, we quantified the shift of the diastolic pressure-volume relation with the EDPVR at the common volume observed in both coronary stenosis and pacing-induced ischemia.
To quantitatively assess the shift of the EDPVR from coronary stenosis to pacing-induced ischemia, we applied two indices, the upward (or downward) shift index and the rightward shift index. As the upward (or downward) shift index, we used end-diastolic pressure during pacing-induced ischemia minus the pressure during coronary stenosis, SI-S, at the largest end-diastolic volume common to both conditions (Fig 4⇓). The two end-diastolic pressures were estimated from equations fitted to the EDPVR. Because end-diastolic volume generally increased from coronary stenosis to pacing-induced ischemia, the largest common volume was usually the maximum end-diastolic volume during coronary stenosis.
As the rightward shift index, we used the largest end-diastolic volume during pacing-induced ischemia minus the largest volume during coronary stenosis, ΔEDVI-S, normalized by left ventricular weight:
where ΔEDVI-S is the rightward shift index, EDVI is the largest end-diastolic volume during pacing-induced ischemia, EDVS is the largest end-diastolic volume during coronary stenosis, and LVW is left ventricular weight.
To examine the effect of each intervention, especially pacing-induced ischemia, on hemodynamic variables, one-way repeated-measures ANOVA and a subsequent Student-Newman-Keuls multiple comparison test were used.
To identify the determinants of the shift of the EDPVR during pacing-induced ischemia, the relation between the upward and rightward shift indices (SI-S and ΔEDVI-S), taken individually, and hemodynamic variables was examined by forward stepwise linear regression with Fenter=4.0 and Fremove=3.9. To select candidate independent variables for the stepwise regression, we first ran bivariate linear regression between the shift index (SI-S and ΔEDVI-S) and each hemodynamic variable and selected variables for which P<.1. Data from the 46 pacing-induced experiments were analyzed together, without including any dummy variables or other considerations for the repeated-measures aspect of the experiment other than normalizing the data for left ventricular weight, as described earlier in “Methods.”
Results are expressed as mean±SD for descriptive statistics and estimate±SEE for regression parameters. Computations were done with sigmastat (Jandel Scientific). Differences were considered to be significant when P<.05.
Hemodynamic Changes in Each Intervention
In 46 experiments, coronary flow (sum of LAD flow and LCx flow) was reduced by about 30% from baseline to coronary stenosis and did not change afterward (Table 1⇓). There were small but significant changes in dP/dtmax, dP/dtmin, systolic segmental shortening of the anterior wall, and the time constant of the isovolumic relaxation from baseline to coronary stenosis but no significant changes in other systolic and diastolic function indices. The small but significant changes in dP/dtmax, dP/dtmin, and systolic segment shortening may represent evidence of some ischemia, albeit mild and probably nontransmural. These results indicate that coronary stenosis (without rapid pacing) did not induce ischemia. In contrast, all left ventricular systolic function variables were significantly impaired and left ventricular relaxation was slowed from coronary stenosis to pacing-induced ischemia, indicating that 3 minutes of rapid pacing imposed on coronary stenosis made the left ventricle ischemic.
Determinants of the Shift of the EDPVR
In the 46 pacing-induced ischemia experiments, the upward shift index, SI-S, was 0.3±1.6 mm Hg (range, −2.0 to 6.4 mm Hg), and the rightward shift index, ΔEDVI-S, was 7.1±4.5 mL/100 g (range, −0.8 to 16.9 mL/100 g). There was an inverse linear correlation between SI-S and ΔEDVI-S (r=−.436, P=.0024), indicating that the more it shifted downward, the more EDPVR shifted rightward (Fig 5⇓).
SI-S had a correlation yielding P<.1 with 12 variables in bivariate linear regression (Table 2⇓), most of which were indices of left ventricular systolic performance or coronary flow. Because percent systolic shortening of segment length was obtained in only 38 experiments, the other 11 variables were put in the stepwise multiple forward regression analysis. The stepwise regression analysis revealed that SI-S was significantly correlated with left ventricular end-diastolic pressure during coronary stenosis, dE/dtmax during pacing-induced ischemia, and Ees during pacing-induced ischemia (Table 2⇓).
ΔEDVI-S had a correlation yielding P<.1 with 22 variables in bivariate linear regression and was significantly correlated with Ees during pacing-induced ischemia, dP/dtmin during pacing-induced ischemia, and stroke volume during coronary stenosis by stepwise regression (Table 3⇓).
As these stepwise regressions showed, the indices of left ventricular contractility, Ees or dE/dtmax, during pacing-induced ischemia were the best determinant, as indicated by the highest partial F value. (In the regression of SI-S, the partial F value was higher for left ventricular end-diastolic pressure during coronary stenosis than dE/dtmax and Ees during pacing-induced ischemia because both indices were selected in the regression. Had only one of the contractility indices during pacing-induced ischemia been selected, the F value associated with this one index would have been even more significant.) Specifically, the significant coefficient of Ees or dE/dtmax during pacing-induced ischemia in the regressions signifies that SI-S decreased and ΔEDVI-S increased as Ees or dE/dtmax decreased.
Determinants of Left Ventricular Contractility
Because left ventricular contractility during pacing-induced ischemia was the best determinant of the shift of the EDPVR, we examined which variables during coronary stenosis or rapid pacing determined left ventricular contractility (Ees and dE/dtmax) during pacing-induced ischemia by forward stepwise regression. Of the 8 (for Ees) and 12 (for dE/dtmax) candidate independent variables (listed in Table 4⇓), coronary flow during rapid pacing was selected as the best determinant of both Ees and dE/dtmax during pacing-induced ischemia (Table 4⇓). Specifically, the more the coronary flow decreased, the more Ees and dE/dtmax decreased.
This study in dogs shows that left ventricular contractility is the best determinant of the shift of the EDPVR during pacing-induced ischemia. In our study, left ventricular end-systolic elastance (Ees) or the slope of the dP/dtmax-Ved relation (dE/dtmax), both indicators of left ventricular contractility, during pacing-induced ischemia was the best determinant of the upward (or downward) shift index (SI-S) and the rightward shift index (ΔEDVI-S) and thus of the shift of the EDPVR. SI-S decreased and ΔEDVI-S increased as Ees or dE/dtmax decreased. Thus, the more left ventricular contractility decreased, the more the EDPVR shifted downward and rightward during pacing-induced ischemia.
Our results are consistent with the results of previous studies that showed that left ventricular systolic function is related to the shift of the diastolic pressure-volume or pressure-length relation during ischemia.4 32 Paulus et al4 observed an inverse correlation between the decrease in segmental stroke work and their index of the upward shift of the diastolic pressure-segment length relation during pacing-induced ischemia in dogs. More recently, Bronzwaer et al32 observed a positive correlation between percent shortening of an ischemic segment and their index of the upward shift of the diastolic pressure–radial length relation during coronary occlusion in humans.
While our result is consistent with these earlier studies, our study provides stronger evidence for a relation between systolic function and the shift of the diastolic pressure-volume relation, for two reasons. First, whereas the earlier studies compared only pressure-volume or pressure-length loops, we compared the EDPVR during vena caval occlusion. Comparing diastolic pressure-volume or pressure-length loops may overestimate the shift when end-diastolic volume or length increases during ischemia, because pressure-volume or pressure-length loop can be affected by an ischemia-induced decrease in the rate of left ventricular relaxation, which results in elevated early diastolic pressure.33 Therefore, the slope of EDPVR is steeper than the slope of diastolic pressure-volume loop during ischemia18 ; a small increase in end-diastolic volume greatly increases end-diastolic pressure. In contrast, comparing EDPVRs allowed us to accurately measure the shift at the same volume before and after ischemia. Second, whereas the earlier studies used bivariate regression analysis, we used multiple regression analysis, which allowed us to assess many potentially important hemodynamic variables rather than to assume a priori that left ventricular contractility is an important determinant of the shift of the EDPVR during pacing-induced ischemia. Thus, the close relation that was demonstrated between left ventricular contractility and the shift of the EDPVR during pacing-induced ischemia was an objective result rather than an assumption of the analysis.
There is a difference in the response of the systolic function when the upward shift of the EDPVR was observed during ischemia between the in situ heart in dogs or humans and the isolated heart in rabbits or ferrets. In the in situ dog or human heart, the systolic function was only mildly depressed (10% to 20% decrease in systolic pressure) when the upward shift of the diastolic pressure-volume relation was observed.1 2 3 4 5 10 32 34 In contrast, in the isolated rabbit or ferret heart, the upward shift has been observed while systolic function was severely impaired (50% decrease or more in systolic pressure).35 36 37 38 The difference in reduction of systolic function when an upward shift of the diastolic pressure-volume relation is observed might be due to the difference in the response of left ventricular relaxation to ischemia between species. In previous studies that used the in situ dog or human heart, the time constant of left ventricular relaxation during ischemia was from 40 to 70 ms, which is not slow enough to affect end-diastolic pressure.1 2 4 5 10 32 In contrast, the time constant of left ventricular relaxation in the rabbit heart increased from between 30 and 50 (during baseline) to between 150 and 225 ms (during ischemia), which is large enough to affect the end-diastolic pressure at the heart rate of 2 to 4 Hz,35 37 because the diastolic interval is short compared with the time constant of isovolumic relaxation. (A study that used ferret heart38 did not report the time constant of isovolumic relaxation, but traces of left ventricular pressure decay appear to be much slower than those in dog or human.) Thus, the role of left ventricular relaxation on the EDPVR during ischemia in rabbit or ferret heart appears to be much more important than in dog or human heart.
The observation that the shift of the EDPVR had various directions and magnitudes during pacing-induced ischemia suggests that the distinction between supply and demand ischemia may not adequately discriminate between the shifts of the diastolic pressure-volume relation. This fact also points to the difficulty of creating the pacing ischemia model, which has not been previously reported. Although an upward shift of the diastolic pressure-volume or pressure–segment length relation has been the hallmark during pacing-induced (demand) ischemia and a rightward or downward shift has been the hallmark during coronary occlusion (supply ischemia),1 2 4 5 6 7 recent studies have observed that the relation can shift in any direction during both types of ischemia.8 9 10 34 From these recent observations, Applegate et al8 suggested that the response of the diastolic pressure-volume relation to ischemia does not depend simply on whether supply or demand ischemia is present but on the balance between supply to and demand of the myocardium. The present study strongly supports their suggestion; various directions and magnitudes of the shift of the EDPVR were observed during one type of ischemia (pacing-induced ischemia) (Fig 2⇑), and furthermore, coronary flow (supply to the heart) was correlated with left ventricular contractility (Table 4⇑), which was the best determinant of the shift of the EDPVR during pacing-induced ischemia.
Pathophysiological mechanisms that could explain the findings of the present study have been proposed by investigators who tried to explain the comparative effects of demand and supply ischemia on the shift of the diastolic pressure–segment length relation.4 5 39 According to these investigators’ proposal, during demand ischemia, myoplasmic calcium increases because of impaired calcium uptake by the sarcoplasmic reticulum. The increase in diastolic myoplasmic calcium leads to persistent (from systole) cross-bridge interaction of the myofilaments during diastole, which leads to the upward shift of the diastolic pressure-volume relation.40 In contrast, during supply ischemia, although myoplasmic calcium increases further, this increase is accompanied by a buildup of tissue metabolites such as H+ and inorganic phosphate due to decreased washout by coronary flow. These metabolites, as well as intracellular acidosis (induced by accumulated H+), inhibit contractile activity by reducing the calcium sensitivity of the myofilaments and thus induce contractile failure despite higher myoplasmic calcium. The diastolic pressure-volume relation cannot shift upward in this situation because the myofilaments cannot interact during diastole if they do not interact during systole.41
Although we do not think that there is a clear distinction between demand and supply ischemia for the reasons described above, our findings could be explained by these pathophysiological mechanisms. Our finding that left ventricular contractility was correlated with coronary flow could be explained if decreased coronary flow increases production of metabolites, decreases their washout, and thus reduces calcium sensitivity of the myofilaments. The finding that the shift of the EDPVR was correlated with left ventricular contractility could also be explained if decreased contractility leads to weak systolic interaction of the myofilaments and results in weak interaction during diastole.
Other possible mechanisms, such as coronary vascular turgor,42 right ventricular interaction,9 10 43 and pericardial constraint,10 18 44 which have also been proposed to explain the upward shift of the diastolic pressure-volume relation during ischemia or comparative effects of the two types of ischemia on the relation, seem less important in explaining our findings. Coronary vascular turgor and coronary arterial pressure, a determinant of coronary vascular turgor, have been reported to influence left ventricular systolic performance45 and left ventricular diastolic distensibility.46 47 48 However, our results showed that the shift of the EDPVR was much better correlated with left ventricular contractility than with coronary flow (bivariate analysis in Tables 2⇑ and 3⇑). As the analysis of the determinants of left ventricular contractility showed, coronary flow during rapid pacing may determine left ventricular contractility during pacing-induced ischemia, but the direct effect of coronary flow (coronary vascular turgor) on the shift of the EDPVR appears weaker than the effect of contractility. In addition, several previous studies showed that rapid pacing in the absence of coronary stenosis, in which coronary flow should have been much higher than in the presence of coronary stenosis, did not induce any shift of the diastolic pressure-volume or pressure-length relation.1 5 8 12 Thus, the existence of some degree of ischemia seems necessary to observe the shift, and coronary vascular turgor does not seem to explain the shift of the EDPVR during pacing-induced ischemia.
Right ventricular interaction also seems unlikely to explain our findings, because right ventricular pressure did not predict the shift of the EDPVR. Pericardial constraint has been proposed to explain the upward shift of the diastolic pressure-volume loops10 18 (not the EDPVR) during coronary occlusion. (Applegate,18 who investigated the effect of pericardium on the diastolic pressure-volume relation during coronary occlusion, reported that although diastolic pressure-volume loops exhibited substantial upward shift when the pericardium was intact and the baseline ventricular volume was large, there was no apparent shift in the EDPVR.) However, since our experiment was done in the open-pericardium heart, pericardial constraint does not explain our results.
Slowed relaxation of the left ventricle is also a possible mechanism of the upward shift of the diastolic pressure-volume relation during ischemia.49 50 Udelson et al50 reported that in patients with hypertrophic cardiomyopathy, which is known to slow left ventricular relaxation, atrial rapid pacing (120 to 135 bpm) shifted the pressure-volume relation upward, but isoproterenol improved the slowed relaxation and attenuated the upward shift despite more severe myocardial ischemia compared with atrial pacing. Their data appear to be inconsistent with our results, because isoproterenol-induced ischemia potentiated left ventricular performance but still attenuated the upward shift. However, in our study, which used intact hearts, the rate of left ventricular relaxation was faster than one in the study by Udelson et al, and the heart rate was slower in their study. In our study, the time constant of left ventricular relaxation was slowed during pacing-induced ischemia but was still not slow enough to affect the end-diastolic pressure, indicating that slowed relaxation does not affect end-diastolic pressure. Thus, the difference in left ventricular relaxation may explain the difference between the results of Udelson et al and ours.
Three potential limitations of the study need to be considered. One potential limitation is that, except for normalizing for left ventricular size, our statistical analysis did not explicitly model between-dog variability. We could not explicitly account for between-dog effects because we did not collect our data in a completely balanced, repeated-measures design. Rather, we used data collected as part of a study that sought to make an upward shift of the diastolic pressure-volume relation during pacing-induced ischemia to see whether gadolinium blocks the upward shift11 ; we did not try to induce a rightward or downward shift in the dogs in which we obtained an upward shift. In addition, we failed to create a large upward shift in 10 of the 15 dogs.
To weight the data for each dog equally, we conducted an additional analysis in which only one multivariate datum was used from each dog. We calculated the mean values for all experiments performed in each dog and examined the relation between the mean shift index values (SI-S and ΔEDVI-S) and the mean hemodynamic variables by linear regression for the 15 points. In this analysis, in which each dog contributed only one data point, dE/dtmax during pacing-induced ischemia was the most significant variable that best correlated with both SI-S and ΔEDVI-S (Table 5⇓ and Fig 6⇓). Ees, another index of contractility, during pacing-induced ischemia was also significantly correlated with ΔEDVI-S and borderline (r=.46, P=.09) correlated with SI-S. These results obtained in additional analysis were consistent with those of the analysis that used 46 pacing-induced ischemia experiments described in “Results.”
The second potential limitation of the study is that we did not account for the possible preconditioning effect on the EDPVR. Ischemic preconditioning, defined as one or more brief episodes of ischemia (usually induced by coronary occlusion or aortic cross-clamping), has been known to increase myocardial tolerance to subsequent sustained ischemia.51 52 Because a number of pacing-induced ischemia experiments were done in some dogs, it could be that the shift of the EDPVR was affected by the preconditioning effect. However, careful review of our data revealed no relation between the shift indices and the sequence number of pacing-induced ischemia experiments in each dog. In addition, in a control study in which pacing-induced ischemia was induced twice in succession by the same extent of coronary stenosis and the same pacing rate,11 the response of the end-diastolic pressure-volume points to repeated pacing-induced ischemia was reproducible. This result suggests that there was no preconditioning effect by a single episode of pacing-induced ischemia. Furthermore, in one dog (dog 452) in which 9 pacing-induced ischemia experiments were performed and various shifts of the EDPVR were observed, there was a significant correlation between SI-S and Ees during pacing-induced ischemia (r=.69, P=.037) and between ΔEDVI-S and Ees during pacing-induced ischemia (r=−.83, P=.005). Thus, in our model of pacing-induced ischemia, a preconditioning effect does not seem to have significantly affected our results that left ventricular contractility determined the shift of the EDPVR.
The third important limitation of this study is that we did not alter left ventricular contractility independently of the severity of ischemia, but we changed the severity of ischemia by altering the extent of coronary stenosis and the rate of rapid pacing. Thus, one may argue that our conclusion that left ventricular contractility determines the shift of the EDPVR is doubtful. In fact, it is reasonable to consider that both contractility and the EDPVR had already been determined at the end of rapid pacing. Thus, from the point of view of time, it might be better to conclude that the severity of ischemia determines the shift of the EDPVR during pacing-induced ischemia. As described above, rapid pacing in the absence of coronary stenosis does not induce the shift of the EDPVR. We also observed minimal shift in some of the pacing-induced ischemia experiments (minimal shift in Fig 3⇑), in which we assumed that coronary stenosis was not severe enough to induce ischemia. As the severity of ischemia increases, the EDPVR would shift upward, rightward, and finally downward. However, as our results suggested, we believe that the severity of ischemia determined left ventricular contractility, which, in turn, determined the shift of the EDPVR. Nevertheless, this is certainly a limitation in the design of our study, although it would be quite difficult to independently vary contractility at a constant severity of ischemia.
In conclusion, this study suggests that left ventricular contractility is the best determinant of the shift of the left ventricular EDPVR during pacing-induced ischemia. The more left ventricular contractility decreases, the more the EDPVR shifts downward and rightward.
This work was supported by NIH grant HL-25869. We thank James Stoughton for technical assistance; Steve Sizemore, Victor Hargrave, and Ashutosh Goel for help with the computers; and David Teitel and Mimi Zeiger for helpful comments on drafts of the manuscript.
- Received August 26, 1994.
- Revision received November 14, 1994.
- Accepted November 26, 1994.
- Copyright © 1995 by American Heart Association
Serizawa T, Carabello BA, Grossman W. Effect of pacing-induced ischemia on left ventricular diastolic pressure-volume relations in dogs with coronary stenoses. Circ Res. 1980;46:430-439.
Paulus WJ, Serizawa T, Grossman W. Altered left ventricular diastolic properties during pacing-induced ischemia in dogs with coronary stenosis: potentiation of caffeine. Circ Res. 1982;50:218-227.
Bourdillon PD, Paulus WJ, Serizawa T, Grossman W. Effects of verapamil on regional myocardial diastolic function in pacing-induced ischemia in dogs. Am J Physiol. 1986;251:H834-H840.
Paulus WJ, Grossman W, Serizawa T, Bourdillon PD, Pasipoularides A, Mirsky I. Different effects of two types of ischemia on myocardial systolic and diastolic function. Am J Physiol. 1985;248:H719-H728.
Momomura S, Ingwall JS, Parker JA, Sahagian P, Ferguson JJ, Grossman W. The relationship of high energy phosphate, tissue pH, and regional blood flow to diastolic distensibility in the ischemic dog myocardium. Circ Res. 1985;57:822-835.
Applegate RJ, Walsh RA, O’Rourke RA. Comparative effects of pacing-induced ischemia and flow-limited ischemia on left ventricular function. Circulation. 1990;81:1380-1392.
Wijns W, Serruys PW, Slager CJ, Grimm J, Krayenbuehl HP, Hugenholtz PG, Hess OM. Effect of coronary occlusion during percutaneous transluminal angioplasty in humans on left ventricular chamber stiffness and regional diastolic pressure-radius relations. J Am Coll Cardiol. 1986;7:455-463.
Kass DA, Midei M, Brinker J, Maughan WL. Influence of coronary occlusion during PTCA on end-systolic and end-diastolic pressure-volume relations in humans. Circulation. 1990;81:447-460.
Takano H, Glantz SA. Effect of gadolinium on the upward shift of the left ventricular diastolic pressure-volume relation during pacing-induced ischemia in dogs. Circulation. In press.
Shintani H, Glantz SA. Influence of filling on left ventricular diastolic pressure-volume curve during pacing ischemia in dogs. Am J Physiol. 1994;266:H1373-H1385.
Boltwood CMJ, Appleyard RF, Glantz SA. Left ventricular volume measurement by conductance catheter in intact dogs: parallel conductance volume depends on left ventricular size. Circulation. 1989;80:1360-1377.
Shintani H, Glantz S. Effects of disrupting the mitral apparatus on left ventricular diastolic properties and contractility in dogs. Circulation. 1993;87:2001-2015.
Kass DA, Marino P, Maughan WL, Sagawa K. Determinants of end-systolic pressure-volume relations during acute regional ischemia in situ. Circulation. 1989;80:1783-1794.
Applegate RJ. Load dependence of left ventricular diastolic pressure-volume relations during short-term coronary artery occlusion. Circulation. 1991;83:661-673.
Kono A, Maughan WL, Sunagawa K, Hamilton K, Sagawa K, Weisfeldt ML. The use of left ventricular end-ejection pressure and peak pressure in the estimation of the end-systolic pressure-volume relationship. Circulation. 1984;70:1057-1065.
Weiss JL, Frederiksen JW, Weisfeldt ML. Hemodynamic determinants of the time-course of fall in canine left ventricular pressure. J Clin Invest. 1976;58:751-760.
Raff GL, Glantz SA. Volume loading slows left ventricular isovolumic relaxation rate: evidence of load-dependent relaxation in the intact dog heart. Circ Res. 1981;48:813-824.
Ohtani M, Nikolic S, Glantz SA. A new approach to in situ left ventricular volume clamping in dogs. Am J Physiol. 1991;261:H1335-H1343.
Suga H, Sagawa K, Shoukas AA. Load dependence of the instantaneous pressure-volume ratio of canine left ventricle and effects of epinephrine and heart rate on the ratio. Circ Res. 1973;32:314-322.
Burkhoff D, Sugiura S, Yue DT, Sagawa K. Contractility-dependent curvilinearity of end-systolic pressure-volume relations. Am J Physiol. 1987;252:H1218-H1227.
van der Velde ET, Burkhoff D, Steendijk P, Karsdon J, Sagawa K, Baan J. Nonlinearity and load sensitivity of end-systolic pressure-volume relation of canine left ventricle in vivo. Circulation. 1991;83:315-327.
Little WC, Cheng CP, Peterson T, Vinten-Johansen J. Response of the left ventricular end-systolic pressure-volume relation in conscious dogs to a wide range of contractile states. Circulation. 1988;78:736-745.
Kass DA, Beyar R, Lankford E, Heard M, Maughan WL, Sagawa K. Influence of contractile state on curvilinearity of in situ end-systolic pressure-volume relations. Circulation. 1989;79:167-178.
Mirsky I. An appraisal of ventricular and myocardial function variables based on elastance concept. J Am Coll Cardiol. 1989;14:354-356.
Little WC. The left ventricular dP/dtmax-end-diastolic volume relation in closed-chest dogs. Circ Res. 1985;56:808-815.
Hess OM, Schneider J, Koch R, Bamert C, Grimm J, Krayenbuehl HP. Diastolic function and myocardial structure in patients with myocardial hypertrophy: special reference to normalized viscoelastic data. Circulation. 1981;63:360-371.
Glantz SA. Ventricular pressure-volume curve indices change with end-diastolic pressure. Circ Res. 1976;39:772-778.
Bronzwaer JGF, Bruyne B, Ascoop CAPL, Paulus WJ. Comparative effects of pacing-induced and balloon coronary occlusion ischemia on left ventricular diastolic function in man. Circulation. 1991;84:211-222.
Carroll JD, Hess OM, Hirzel HO, Krayenbuehl HP. Exercise-induced ischemia: the influence of altered relaxation on early diastolic pressures. Circulation. 1983;67:521-528.
Isoyama S, Apstein CS, Wexler LF, Grice WN, Lorell BH. Acute decrease in left ventricular diastolic chamber distensibility during simulated angina in isolated hearts. Circ Res. 1987;61:925-933.
Lorell BH, Isoyama S, Grice WN, Weinberg EO, Apstein CS. Effects of ouabain and isoproterenol on left ventricular diastolic function during low-flow ischemia in isolated, blood-perfused rabbit hearts. Circ Res. 1988;63:457-467.
Serizawa R, Vogel WM, Apstein CS, Grossman W. Comparison of acute alterations in left ventricular relaxation and diastolic chamber stiffness induced by hypoxia and ischemia. J Clin Invest. 1981;68:91-102.
Kihara Y, Grossman W, Morgan JP. Direct measurement of changes in intracellular calcium transients during hypoxia, ischemia, and reperfusion of the intact mammalian heart. Circ Res. 1989;65:1029-1044.
Paulus WJ. Upward shift and outward bulge: divergent myocardial effects of pacing angina and brief coronary occlusion. Circulation. 1990;81:1436-1439.
De Bruyne B, Bronzwaer JGF, Heyndricks GR, Paulus WJ. Comparative effects of ischemia and hypoxemia on left ventricular systolic and diastolic function in humans. Circulation. 1993;88:461-471.
Hess OM, Osakada G, Lavelle JF, Gallagher KP, Kemper WS, Ross JJ. Diastolic myocardial wall stiffness and ventricular relaxation during partial and complete coronary occlusions in the conscious dog. Circ Res. 1983;52:387-400.
Gregg DE. Effect of coronary perfusion pressure or coronary flow on oxygen usage of the myocardium. Circ Res. 1963;13:497-500.
Salisbury PF, Cross CE, Rieben PA. Influence of coronary artery pressure upon myocardial elasticity. Circ Res. 1960;8:794-800.
Gaasch WH, Bing OHL, Franklin A, Rhodes D, Bernard SA, Weintraub RM. The influence of acute alterations in coronary blood flow on left ventricular diastolic compliance and wall thickness. Eur J Cardiol. 1978;7(suppl 1):147-161.
Vogel WM, Apstein CS, Briggs LL, Gaasch WH, Ahn J. Acute alterations in left ventricular stiffness: role of the erectile effects of coronary artery pressure and flow in normal damaged hearts. Circ Res. 1982;51:465-478.
Gilbert JC, Glantz SA. Determinants of left ventricular filling and of the diastolic pressure-volume relation. Circ Res. 1989;64:827-852.
Udelson JE, Cannon RO III, Bacharach SL, Rumble TF, Bonow RO. β-Adrenergic stimulation with isoproterenol enhances left ventricular diastolic performance in hypertrophic cardiomyopathy despite potentiation of myocardial ischemia: comparison to rapid atrial pacing. Circulation. 1989;79:371-382.
Murry CE, Jennings RB, Reimer KA. Preconditioning with ischemia: a delay of lethal cell injury in ischemic myocardium. Circulation. 1986;74:1124-1136.
Cohen MV, Liu GS, Downey JM. Preconditioning causes improved wall motion as well as smaller infarcts after transient coronary occlusion in rabbits. Circulation. 1991;84:341-349.