INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES APPENDIX

THE BLOOD PRESENT in the entire coronary circulation (arteries, arterioles, capillaries, venules, and veins) is ~12 ml/100 g of cardiac muscle (17). Although generally called myocardial blood volume, the correct term for this entity should be coronary blood volume (19). Approximately one-third of this blood is present within the myocardium itself, which can correctly be termed the myocardial blood volume (19). Close to 90% of the myocardial blood volume resides in capillaries (17), and could be called the capillary blood volume (capBV). Myocardial contrast echocardiography (MCE) utilizes the venous administration of microbubbles of the size of red blood cells during simultaneous ultrasound imaging (18). The myocardial signal measured on MCE reflects the concentration of microbubbles within the myocardium. During a continuous infusion of microbubbles, when the entire ultrasound beam is fully replenished with microbubbles, the increased signal from the myocardium largely denotes the capBV (19, 23, 29, 30). Any alteration in myocardial signal in this situation must, therefore, occur principally due to a change in capBV.

Using a continuous infusion of microbubbles during MCE, we have previously shown that in the absence of a stenosis coronary hyperemia results in an increase in myocardial microbubble velocity which occurs in proportion to increases in coronary blood flow (CBF) (30), without a change in capBV and hence myocardial signal amplitude. That myocardial blood volume does not change during exogenous hyperemia in the absence of stenosis has also been reported using intravital microscopy (27) and other ex vivo methods (6). We have also previously demonstrated that during coronary hyperemia, unlike in the absence of stenosis, myocardial signal on MCE (hence, capBV) decreases distal to a stenosis (23, 29, 30). Additionally, CBF reserve within a myocardial bed supplied by a stenotic coronary artery can be accurately assessed using measurements of MCE-derived ultrasound signal in that bed relative to a normal bed (30).

On the basis of the findings above, we postulated that the resistance offered by the capillaries is the major factor responsible for limiting CBF during hyperemia, both in the absence and presence of a noncritical stenosis. We also postulated that the reduction in CBF at rest caused by a severe stenosis would be associated with an increase in capillary resistance. Since capillaries are placed in parallel, any change in their resistance results in a change in their volume (2). Thus capBV would also decrease. Our unifying hypothesis therefore was that, contrary to currently held beliefs, capillaries participate in the regulation of CBF.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES APPENDIX

Model development. We used the following two assumptions while developing our model: 1) CBF is regulated via adjustments of the resistance of different components of the coronary circulation, and 2) the principal reason for autoregulation is maintenance of homeostasis through a constant capillary hydrostatic pressure (15). We modeled the coronary circulation as comprising three vascular beds: arterial, capillary, and venous, connected in series (Fig. 1), each with its own resistance (R_a, R_c, and R_v, respectively). Total coronary resistance was obtained by dividing the coronary driving pressure (P), across the entire coronary bed by the CBF (Q) to that bed.

View larger version (26K): [in this window] [in a new window]   Fig. 1.   Schema of experimental preparation and the locations of the 3 vascular components whose resistance is calculated by the model. Rs, Ra, Rc, Rv, stenosis, arterial, capillary, and venous resistances, respectively. See text for details.

At baseline (in the absence of both hyperemia and stenosis)

(1)

and

(2)

where P_ra is the mean right atrial pressure, and 30 is the mean pressure (mmHg) measured in the center of the capillary bed (28).

The placement of a noncritical stenosis at rest produces no change in capBV (23, 29, 30), implying that R_c remains constant. Any change in total microvascular resistance, therefore, will occur from changes in arteriolar and/or venular resistances. In this setting, P will be reduced to P_s, and R_a and R_v will be reduced to R_as and R_vs, respectively. Therefore, Eqs. 1 and 2 can be rewritten as

(3)

and

(4)

where Q_s is the CBF during stenosis.

When hyperemia is induced in the absence of a stenosis, myocardial blood volume does not change (6, 27, 29, 30), which means that R_c does not change. We can therefore assume that any changes in total microvascular resistance will occur from changes in resistances of arterioles and/or venules that are not present within the myocardium itself. In this setting, R_a and R_v will be reduced to R_ah and R_vh, respectively, and Eqs. 1 and 2 can be rewritten as

(5)

and

(6)

where P_h and Q_h are the coronary driving pressure and CBF during hyperemia, respectively.

When a stenosis is applied during coronary hyperemia, P_h decreases to P_hs. On the basis of our previous observations of decrease in capBV during hyperemia in the presence of stenosis (23, 29, 30), we can assume that R_c increases to R_cs. We can also assume that R_ah and R_vh are already minimal because of hyperemia and do not decrease any further in the presence of a stenosis. We can, therefore, rewrite Eqs. 1 and 2 as

(7)

and

(8)

where P_hs is coronary driving pressure, and Q_hs is CBF during stenosis and hyperemia.

These eight simultaneous equations with eight unknown parameters can be written in matrix form as

and solved using matrix inversion by singular value decomposition.

Animal preparation. The study protocol was approved by the Animal Research Committee at the University of Virginia and conformed to the American Heart Association Guidelines for Use of Animals in Research. Nine adult mongrel dogs were used for the study. They were anesthetized with 30 mg/kg pentobarbital sodium (Abbott), intubated, and ventilated with a respirator pump (model 607, Harvard Apparatus). Additional anesthesia was administered as needed. Catheters (7-Fr.) were placed in the femoral veins for infusion of fluids and drugs and for the administration of ultrasound contrast agent for MCE.

A left lateral thoracotomy was performed, and the heart was suspended in a pericardial cradle. A 7-Fr. fluid-filled catheter was placed in the ascending aorta via the right carotid artery for the measurement of aortic pressure (Fig. 1). A 7-Fr. micromanometer-tipped catheter (Millar Instruments) was positioned in the right atrium via the internal jugular vein for the measurement of right atrial pressure (Fig. 1). Proximal portions of the left anterior descending (LAD) and left circumflex (LCx) arteries were dissected free from the surrounding tissue. Ultrasonic time-of-flight flow probes (series SB, Transonic) were placed around both vessels and connected to a digital flowmeter (model T206, Transonic). A fluid-filled 20-gauge Teflon catheter was introduced into the LAD via a side branch to measure distal coronary pressure, and a custom-designed screw occluder was placed proximal to the tip of this catheter to produce coronary stenoses of varying severities (Fig. 1).

Hemodynamic measurements. The fluid-filled aortic and LAD catheters were connected to a fluid-filled pressure transducer, which, like the micromanometer-tipped catheter and the flowmeter, were connected to a multichannel recorder (model ES 2000, Gould Electronics). The fluid-filled transducer was carefully calibrated before each experiment to zero pressure by opening to room air, and to 200 mmHg pressure by using an air pump (X-caliber System, Gould Electronics). The micromanometer-tipped catheter was similarly calibrated. Epicardial CBF and all pressures were acquired digitally at 200 Hz into a computer via an 8-channel analog-to-digital converter (Metrabyte) (22). The signals were displayed online using Labtech Notebook (Laboratory Technologies). The severity of each stenosis was judged by its resistance (R_s), which was calculated by dividing the mean gradient between the aortic and the distal LAD pressure by CBF (Q). Coronary driving pressure (P) was calculated by subtracting P_ra from the mean distal LAD pressure, and resistance across the myocardial microvasculature was calculated as P/Q.

Myocardial contrast echocardiography. Intermittent harmonic imaging was performed with a Sonos 5500 system (Hewlett-Packard), where ultrasound was transmitted at 2.0 MHz and received at 4.0 MHz (23, 29, 30). The transducer was fixed in position with a custom-designed clamp attached to the procedure table, and a saline bath served as an acoustic interface between the transducer and the heart. The maximal dynamic range (60 dB) was used. The transmit power, focus, overall gain, and image depth were held constant for all experiments. Data were recorded on 1.25-cm videotape with a S-VHS recorder (Panasonic model AG-MD8300, Matsushita Electrical).

Dual-triggered imaging was used initially to obtain pulsing intervals that were less than a single cardiac cycle (23, 29, 30). The first trigger was used for microbubble destruction, and the second trigger was used to acquire images in end- systole. Subsequently, pulsing intervals ranging from 1 to 20 cardiac cycles were obtained using a single trigger in end systole. The resulting pulsing intervals, ranging from 150 ms to 12 s, allowed progressively greater bubble replenishment of the ultrasound beam elevation. Up to eight end-systolic images were acquired at each pulsing interval.

MCE images were transferred from videotape to an off-line image analysis system in a 340 × 240 × 8-bit format and analyzed as previously described (23, 29, 30). Five images acquired at each of the 10 pulsing intervals were aligned using computer cross correlation, so were 4-6 frames before the appearance of microbubbles in the myocardium, which were considered to represent background. Pulsing interval vs. background-subtracted video intensity (VI) plots were generated from each pixel in the aligned and averaged images and fitted to an exponential function: y = A(1  e^t), where y is VI at a pulsing interval t, A is the plateau VI representing myocardial blood volume, and  is the rate constant reflecting the rate of rise of VI (or the mean myocardial microbubble velocity) (23, 29, 30). Both MCE parameters (A and ) were then displayed as color-coded parametric images (23). Average values of these parameters were derived by placing regions of interest over the central 50% of the LAD and LCx beds, with care taken to avoid the specular endocardial and epicardial borders.

Experimental protocol. First, the LAD bed was defined in each dog by injecting 0.2 ml of Albunex (Molecular Biosystems), diluted with 0.8 ml of normal saline through the catheter placed in it. A constant injection rate of 1 ml/s was used, which produced excellent myocardial opacification and yet avoided reflux of the microbubbles into the aortic root or the LCx (22).

A baseline stage (no stenosis) was first performed at rest and during hyperemia induced by a venous infusion of 0.4 µg · kg¹ · min¹ of WRC-0470 (Medco Research), a selective adenosine A_2a receptor agonist (9). MCE was performed using a solution consisting of 4 ml of Imagent (Alliance Pharmaceutical) in 50 ml of normal saline administered as a continuous infusion via the femoral vein at a rate of 2 ml/min using a volumetric pump (IVAC) (23, 30). This infusion rate allowed adequate myocardial opacification and at the same time minimized the posterior wall attenuation from microbubbles present within the left ventricular cavity.

Up to two stenosis stages were then attempted in each dog, and all measurements were repeated. The stenosis severity was judged by the pressure gradient across it and its effect on resting CBF. In the presence of noncritical stenosis, measurements were also made during hyperemia. In the case of severe stenosis, which resulted in decrease in resting CBF, hyperemia was not induced.

Statistical methods. Data are expressed as means ± 1 SD. Interstage comparisons were performed using the paired Student's t-test. Correlations were performed using least-squares fit linear regression analysis. P < 0.05 (two-sided) was considered statistically significant for all comparisons.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES APPENDIX

Baseline and hyperemia stages without stenosis were obtained in all 9 dogs. In addition, 13 stenosis stages were obtained: 7 noncritical stenoses were obtained in 6 dogs, and 6 stenoses with reduction in resting CBF were obtained in 5 dogs. Hyperemia stages were also obtained for the noncritical stenoses. Thus the total number of stages (including rest and hyperemia) analyzed was 38.

Noncritical stenosis. Table 1 depicts hemodynamic and MCE data during rest at baseline and after placement of noncritical stenosis, which did not result in reduction in resting CBF. A mild gradient was measured between the aorta and distal LAD at baseline, which probably resulted from the screw occluder placed around the LAD and from which the stenosis resistance (R_s) was calculated. As expected, the coronary driving pressure (P) decreased in the presence of a stenosis (by 9 to 34 mmHg), as did the resistance distal to the stenosis. Because myocardial VI (A) did not change between baseline and stenosis, it can be assumed that R_c did not change. Furthermore, since P decreased in the presence of constant CBF, and R_c did not change, it follows that either R_a or R_v, or both, must have decreased. Since CBF was not different between the two stages, it is not surprising that microbubble velocity () and the product A ·  (which represents myocardial blood flow by MCE; see Refs. 23, 29, 30) also remained unchanged.

Table 2 shows the hemodynamic and MCE data before and after placement of stenosis during hyperemia. Although CBF increased during hyperemia in the absence of a stenosis, compared with at rest (Table 1), this increase was attenuated after stenosis placement. Because of hyperemia, the stenosis gradient increased compared with at rest, and the coronary driving pressure decreased. Although hyperemia resulted in a decrease in myocardial vascular resistance compared with at rest (Table 1), the resistance increased after placement of a stenosis. This finding was associated with a decrease in myocardial VI (A), which implies a decrease in capBV and an increase in R_c. In keeping with attenuation of hyperemic CBF after stenosis placement, the microbubble velocity () and the product A ·  decreased.

Figure 2 illustrates parametric MCE images denoting capBV (A) in a dog during hyperemia before (Fig. 2A) and after (Fig. 2, B and C) placement of LAD stenoses. These two stenoses were non-flow limiting at rest. Change in color from yellow, to orange, to red, to no color denotes progressively less capBV. Figure 3 depicts pulsing interval vs. VI curves obtained from the LAD bed during these three stages. capBV (A) and microbubble velocity () are the highest before stenosis and decrease with the severity of stenosis.

View larger version (54K): [in this window] [in a new window]   Fig. 2.   Color-coded parametric images demonstrating capillary blood volume (capBV) during hyperemia before (A) and after placement of stenoses of varying severities. (B and C). Colors yellow, to orange, to red, to no color denote incremental decreases in capBV. See text for details.

View larger version (13K): [in this window] [in a new window]   Fig. 3.   Pulsing interval vs. video intensity (VI) plots obtained during hyperemia before () and after placement of stenoses of different severities ( and ). Data are from the same stages from where images are depicted in Fig. 2. See text for details.

Figure 4 illustrates the mean stenosis as well as the model-derived mean arteriolar, capillary, and venous resistances at rest and hyperemia, both in the absence and presence of stenosis. As expected, in the absence of hyperemia or stenosis, maximal resistance (61 ± 5% of total myocardial vascular resistance) was offered by the arteriolar component of the coronary circulation, with only a small portion offered by the capillary and venous components (25 ± 5% and 14 ± 4%, respectively). When a noncritical stenosis was applied at rest, the total myocardial vascular resistance decreased to 78 ± 10% of baseline, with no change in capillary resistance. The arteriolar resistance decreased by 38 ± 16% and comprised 48 ± 8% of the total myocardial vascular resistance compared with 61 ± 5% in the absence of a stenosis (P < 0.01). The increase in stenosis resistance of 0.44 ± 0.27 mmHg · ml¹ · min was compensated by a decrease (0.66 ± 0.34 mmHg · ml¹ · min) in arteriolar resistance. A negligible increase (0.04 ± 0.10 mmHg · ml¹ · min) in venular resistance was noted. The individual values in each dog were related to the severity of the stenosis.

View larger version (23K): [in this window] [in a new window]   Fig. 4.   Calculated Rs as well as the model-derived Ra, Rc, and Rv are shown at rest (R), during noncritical stenosis at rest (R+S), during hyperemia (H), and during hyperemia and noncritical stenosis (H+S). See text for details.

When hyperemia was induced in the absence of a stenosis, the total myocardial vascular resistance decreased by 68 ± 7%. Compared with rest, the arterial and venular resistances decreased by 86 ± 9% and 98 ± 5%, respectively. Because of a similar decrease in arterial and venular resistances, the capillary hydrostatic pressure remained unchanged. The arteriolar and capillary resistances comprised 24 ± 14% and 75 ± 14% of the total myocardial vascular resistance. Thus capillaries offered the most resistance to CBF during hyperemia, when no stenosis was present.

In the presence of a noncritical stenosis, hyperemia resulted in a net decrease in total myocardial vascular resistance of 51 ± 13% compared with baseline. Of note, however, was the 58 ± 50% increase in the total myocardial vascular resistance compared with the absence of stenosis. During hyperemia, arteriolar and venular resistances were already minimal, so the increase in resistance caused by placement of a noncritical stenosis occurred from a 81 ± 71% increase in capillary resistance, making it the predominant source (84 ± 8%) of resistance distal to a stenosis. Thus the major reason for attenuation of CBF reserve caused by a stenosis was an increase in capillary resistance.

Figure 5 illustrates the relation between stenosis severity and CBF during hyperemia. The percent diameter stenosis was calculated using 1  (R_o/R_s)^1/4 derived from Poiseuille's equation, where R_o is the resistance offered by a normal coronary artery segment 0.5 cm in length with a 3-mm luminal diameter. Data and exponential fits of the form y = A[1  e^(x  x0)] are shown for experimentally observed CBF, as well as for CBF calculated if the total vascular resistance was due to the stenosis alone [using the equation (P_ao  P_ra)/R_s], arterioles alone [using the equation (P_ao  P_ra)/R_ah], and capillaries alone [using the equation (P_ao  P_ra)/R_c], where P_ao is the mean aortic pressure. The experimental data match most closely the curve where the total resistance is offered by the capillaries alone.

View larger version (13K): [in this window] [in a new window]   Fig. 5.   Relation between stenosis severity (x-axis) and coronary blood flow (CBF, y-axis) during hyperemia in dogs with noncritical coronary stenosis. Actual measured CBF is shown as data () fitting a solid line (A). The other curves denote the following: CBF if the entire resistance were due to the stenosis alone ( and dashed line, B), capillaries alone ( and dash-dotted line, C), and arterioles alone ( and dotted line, D). See text for details.

According to Ohm's law, the reciprocal of CBF equals the sum of the reciprocals of CBF due to the resistances of the different coronary vascular components, which can be calculated using the equation R/(P_ao  P_ra). Consequently, the contributions of the resistances of the different coronary vascular components to the total hyperemic CBF is best illustrated by plotting the reciprocal of CBF against the percent diameter stenosis (Fig. 6). The solid line in Fig. 6 shows the reciprocal of the CBF measured experimentally vs. the calculated coronary stenosis. The vertical distance between the upper and lower limits of the shaded area denotes the contribution of capillary resistance to 1/CBF at any given level of stenosis. It is clear that this component contributes the most during hyperemia and is maximal when stenosis severity exceeds 75%. The vertical distance between the lower limit of the shaded line and the x-axis denotes the contribution of the stenosis resistance to 1/CBF. Although this component also increases when stenosis exceeds 75%, its contribution to 1/CBF is markedly less than the contribution from the capillary resistance. Finally, the vertical distance between the upper boundary of the shaded area and the solid line represents the contribution of arteriolar resistance to 1/CBF. Since arterioles are fully dilated during hyperemia, this component is the smallest, and its contribution decreases when stenosis severity exceeds 75%.

View larger version (18K): [in this window] [in a new window]   Fig. 6.   Relation between stenosis severity (x-axis) and the reciprocal of CBF (y-axis). Actual 1/CBF is shown as the solid line. Shaded area denotes the contribution of capillary resistance to 1/CBF at any given level of stenosis. Vertical distance between the lower limit of the shaded area and x-axis denotes the contribution of the stenosis resistance to 1/CBF. Vertical distance between the upper boundary of the shaded area and the solid line represents the contribution of arteriolar resistance to 1/CBF. See text for details.

Figure 7 depicts the relation between the ratio of A (capBV obtained on MCE) and the ratio of the reciprocal of the capillary resistance (model-derived capBV, see APPENDIX) after and before placement of the stenosis. Both linear and quadratic relations fit equally well to the data. Thus it is difficult to determine whether the increase in capillary resistance during hyperemia and stenosis is due to capillary derecruitment (linear relation) or a decrease in capillary diameter (quadratic relation), or both (see APPENDIX).

View larger version (18K): [in this window] [in a new window]   Fig. 7.   Relation of myocardial VI ratio (capBV from MCE) depicted on the x-axis vs. the reciprocal of ratio of capillary resistances (capBV from the model) depicted on the y-axis, during hyperemia after and before placement of a noncritical stenosis. Both linear and quadratic relations fit equally well to the data. See text for details. MCE, myocardial contrast echocardiography. SEE, standard error of the estimate.

Flow-limiting stenosis. Table 3 depicts data from the dogs that underwent placement of severe stenosis resulting in a modest reduction in resting CBF. As expected, this reduction in CBF was associated with a decrease in coronary driving pressure. The total myocardial microvascular resistance, however, did not change despite a flow-limiting stenosis. Myocardial VI decreased, indicating an increase in capillary resistance. In keeping with reduction in CBF,  and the product A ·  were also mildly reduced. Our model could not be used to derive the resistances of the different beds, because during reduction in resting CBF, resistances of all three vascular beds changed, resulting in eight equations with nine unknown variables, which could not be solved.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES APPENDIX

There are three major new findings from this study. First, in the normal situation, increase in CBF during hyperemia is curtailed mainly by capillary resistance. Second, capBV distal to a noncritical stenosis decreases during hyperemia resulting in an increase in capillary resistance, which is the major determinant of reduced CBF reserve. Third, the decrease in resting CBF caused by a severe stenosis is associated with a decrease in capBV, implying that increased capillary resistance, in addition to the stenosis resistance, may be responsible for the decrease in resting CBF. These findings assume that the capillary hydrostatic pressure does not change during hyperemia and that normal fluid mechanics are operational under all conditions studied. Therefore, taken together these findings indicate that, contrary to previously held notions, capillaries play a vital role in CBF regulation.

Changes in vascular resistance at rest in absence and presence of a noncritical stenosis. Figure 8, A and B, depicts pressure changes at rest in the presence and absence of stenosis using a theoretical construct based on our findings. In this example, P_ao and P_ra are assumed to be 90 and 5 mmHg, respectively. At rest (Fig. 8A), P_ao is reduced by half to reach a precapillary pressure of 45 mmHg. The postcapillary pressure is 15 mmHg, making the mean capillary pressure 30 mmHg (28). Since CBF is the same through all compartments, P across each compartment is proportional to the resistance offered by that compartment. The highest resistance is offered by the arterioles and the least by the venules.

View larger version (18K): [in this window] [in a new window]   Fig. 8.   Mechanisms of maintenance of constant capillary hydrostatic pressure at rest in absence of any stenosis (A), at rest in presence of stenosis (B), during hyperemia in absence of any stenosis (C), and during hyperemia in presence of stenosis (D). Dotted lines in the capillary bed in D illustrate decrease in capBV. Numbers are in mmHg. See text for details.

In the presence of a non-flow-limiting stenosis at rest (Fig. 8B), the distal coronary driving pressure drops from 90 to 75 mmHg. The precapillary pressure, however, remains unchanged at 45 mmHg because of the resulting reduction in arteriolar resistance (4, 10, 15, 26), which fully compensates for the resistance offered by the stenosis. The mean capillary pressure is therefore unaffected at 30 mmHg and CBF remains unchanged.

Changes in vascular resistance during hyperemia in absence and presence of a noncritical stenosis. Figure 8C depicts the pressure changes in the coronary vascular compartments during hyperemia in the absence of stenosis, using the same example. In this situation, arteriolar and venular resistances decrease, thus increasing CBF. Since the drop in arteriolar resistance is greater than the increase in CBF, the precapillary pressure increases from 45 to 52 mmHg. In this instance, mean capillary pressure can be maintained constant at 30 mmHg only if the postcapillary pressure drops to 8 mmHg. However, since CBF also increases severalfold with an increase in P across the capillary bed from 30 to 44 mmHg, R_c remains unchanged. Thus, when arteriolar and venular resistances are minimal, capillaries offer the greatest resistance to CBF, which is the major reason why CBF does not increase to higher levels.

In the same example, when a stenosis is applied during hyperemia (Fig. 8D), the distal coronary pressure falls from 90 to 70 mmHg. To prevent precapillary and hence mean capillary pressure from dropping, the pressure gradient across the arterioles must decrease. Because the arteriolar resistance is minimal during hyperemia, this pressure gradient can decrease only by a decrease in hyperemic CBF, which results partially from the resistance offered by the stenosis (10) but (as shown in Figs. 4 to 6) mainly by increased capillary resistance. Since capillary lengths remain virtually constant, the only way in which capillary resistance can increase is by a decrease in capBV (2).

Other than the dimensions, blood viscosity plays a very important role in the increased resistance to CBF noted in microvessels (13, 15, 25). The viscosity term in Poiseuille's equation (see APPENDIX) has a minor role in total resistance offered by large vessels, where the luminal diameter is the most important determinant of resistance. In contradistinction, the viscosity term is the most important determinant of microvascular flow (25). The measured resistance across microvessels is orders of magnitude more than can be explained on the basis of their size alone (13, 14, 25), which implies that they must play a major role in the regulation of CBF.

On superficial examination, our results appear to vary from those of Chilian et al. (4). Their definitions of the vascular compartments, however, were different from ours, and they did not examine the capillary bed specifically. Instead, all microvessels <170 µm in diameter were lumped into a single compartment. Two-thirds of the total coronary vascular resistance at rest was attributed to this compartment, which comprised both high-resistance arterioles as well as capillaries. They also noted a higher resistance offered by the venules during hyperemia. During hyperemia, venules are most likely to decrease their resistance (11). If these are considered to be part of the compartment containing microvessels <170 µm in diameter rather than a separate venular compartment as in our model, then during hyperemia an apparent increase in the resistance offered by the venules outside this compartment will be seen. We believe that these differences in definitions of the vascular compartments explain the apparent discrepancy between their results and ours.

During hyperemia, Chilian et al. (4) noted a decrease in pressure in arterioles >170 µm in diameter and an increase in pressure in venules >170 µm in diameter. The mean capillary pressure should, therefore, have remained constant. However, the function they chose to fit to the data resulted in an apparent increase in capillary pressure, which was not measured but was extrapolated by the function fit (figure 3 in Chilian et al., Ref. 4). A different function could have resulted in no change in capillary pressure. Obviously, the actual changes in capillary pressures can only be derived from in vivo measurements, which are very difficult in the beating heart.

We are unable to determine whether the decrease in capBV distal to a stenosis occurs because of a decrease in capillary dimensions or derecruitment (Fig. 7). It is, however, tempting to speculate the latter as the principal mechanism, since the fit seen in Fig. 7 appears only mildly quadratic. It has been shown that under resting conditions red blood cells pass only through a fraction of open capillaries, presumably because of limited oxygen need (8, 12). Increased oxygen demand causes red blood cells to pass through more capillaries. When the number of capillaries perfused by red blood cells (capBV) decreases without alteration in oxygen demand, as occurs distal to a stenosis during exogenously induced hyperemia, optimal tissue oxygenation is more likely to be achieved when some capillaries are functionally derecruited, while red blood cell transit is maintained through others. Decrease in dimensions of all capillaries may result in slowing of red blood cell transit, which in turn may cause ischemia. Although the mean microbubble velocity during hyperemia decreased after stenosis placement, it did not drop below the resting level prior to stenosis placement. Further direct microscopic studies could help resolve this issue.

Changes in capBV distal to a severe stenosis when resting CBF is reduced. Our results indicate that when a stenosis is sufficiently severe to cause a decrease in resting CBF, capBV distal to it decreases. Although in this setting we could not determine capillary resistance from our model, it is likely that it increases. It has previously been demonstrated that arteriolar vasodilation is not maximal in the presence of severe stenosis, even when CBF reserve is exhausted (3, 5). Thus the increase in stenosis resistance is not fully compensated by a decrease in arteriolar resistance. As a result, the precapillary pressure may drop, which would result in an increase in postcapillary pressure to maintain the mean capillary pressure at a constant level. The resulting decrease in P across the capillary bed is evidently less than the decrease in CBF, and therefore the capillary resistance increases. During low perfusion pressure and CBF, increase in capillary resistance could be achieved through increase in coronary venular resistance. This mechanism has been shown to be operative in the skeletal muscle and is thought to occur from changes in blood rheology in the venules (13-15).

Implications of our findings. Our findings provide novel insights into the physiology of coronary hyperemia and CBF reserve. The presence of epicardial coronary stenoses is considered the most common cause of decreased CBF reserve in the clinical setting. Our results imply that any condition that would affect capBV or blood rheology within capillaries is as likely to affect CBF reserve. Since capBV is markedly reduced in myocardial infarction (18, 19), it should not be surprising that CBF reserve is abnormal in this condition. Other conditions that affect capBV, such as diabetes or hypertension, will similarly affect CBF reserve. Disorders that are associated with alterations in microvascular rheology could also affect CBF reserve. These include the following: endothelial dysfunction causing changes in the thickness, charge, or composition of proteoglycans lining the capillaries; irregularities in luminal contours as well as asymmetry in capillary architecture caused by disease; or alterations in capillary blood cell composition (25).

Our findings also provide an explanation for the critical coronary closing pressure. If the total microvascular resistance is always constant, then the relation between CBF and coronary perfusion pressure should be linear with a zero intercept (shown as a dashed line in Fig. 9). It has, however, been repeatedly demonstrated (7, 20) that the relation does not have a zero intercept (solid line in Fig. 9). One reason for this finding may be that microvascular resistance does not remain constant with changes in CBF. Our results indicate that for any decrease in resting or hyperemic CBF, there is a proportionate decrease in capBV and hence an increase in capillary resistance, resulting in a positive intercept on the x-axis as depicted in Fig. 9. Although other explanations have been evoked for this phenomenon (20), ours is a reasonable alternative explanation.

View larger version (13K): [in this window] [in a new window]   Fig. 9.   Relation between coronary perfusion pressure and CBF. Dashed line depicts the relation when the coronary resistance is constant. Solid line depicts the actually observed relation (nonzero closing pressure). This positive intercept on the x-axis can be explained by the increasing coronary resistance (denoted by the lines from the origin to points A, B, and C) as the perfusion pressure decreases. See text for details.

A clinically important implication of our findings relates to myocardial perfusion imaging. The occurrence of perfusion defects during hyperemia in patients with coronary stenosis is believed to be due to a greater increase in CBF in the normal bed compared with the bed supplied by a stenosis. Because VI on MCE reflects myocardial blood volume and not flow, our results indicate that perfusion defects on MCE occur secondary to a decrease in capBV in the bed supplied by the stenosis. Since capBV is a major determinant of capillary permeability surface area product (1), it follows that uptake of any diffusible tracer will decline distal to a stenosis in the presence of hyperemia. Thus it is conceivable that perfusion defects seen on other imaging modalities may also occur through a decrease in capBV. Similarly, a decrease in capBV may also be responsible for the occurrence of perfusion defects when resting CBF is reduced.

Critique of our methods. Our model does not take collateral CBF into account. We have previously demonstrated that in the acute setting, application of a non-flow-limiting stenosis at rest results in only the peripheral 20% of the perfusion bed size becoming collateral dependent (22). In our model, changes in perfusion bed size are reflected as changes in resistance. In this study, application of a noncritical stenosis reduced the coronary driving pressure by a mean of 24% at rest and 30% during hyperemia. Our results also demonstrate that the stenosis resistance is completely compensated for by a similar decrease in arteriolar resistance, suggesting that collateral flow plays only a minor role in the presence of a noncritical stenosis.

We have previously demonstrated that in the presence of a critical stenosis, when resting CBF is reduced acutely to very low levels (20-30% of normal), no more than 50% of the periphery of the perfusion bed becomes collateral dependent, and the central 50% of the bed is not collateral dependent (22). To ensure that our VI measurements pertain to the capillary bed that is supplied only by the stenotic artery and not by collaterals, the VI measurements were made in the central 50% of the perfusion bed. Any effects of collateral flow on our measurements should, therefore, be minimal. It is, however, possible that the increase in capillary resistance seen at rest during this condition could in part be due to the decrease in perfusion bed size as a result of collateral flow.

We believe that the strength of our model can, in part, be attributed to its simplicity. More complicated models involve several unknown parameters and therefore rely on too many assumptions, many of which are not pertinent to in vivo conditions. Our assumption that mean capillary pressure is maintained at near constant levels during normal to hyperemic CBF is reasonable, based on capillary pressure measurements made in skeletal muscle (15). It is also reasonable to assume that the primary purpose of autoregulation is the maintenance of homeostasis, for which a constant hydrostatic pressure is essential. Making direct measurements of capillary pressures in an in vivo beating heart could confirm this assumption, but it is currently not possible to do so.

The design of our study and the methods used do not permit an evaluation of the proximate mechanism of decrease in capBV distal to a stenosis. Since the changes occur immediately, it has been suggested that they are modulated by myogenic reflexes, particularly in the skeletal muscle (24). It has also been suggested that, at least in the skeletal muscle, viscosity changes in small venules may be responsible for maintaining a constant capillary pressure (13, 14). More intriguing is the putative signal(s) that may initiate changes in capBV and the source of these signals. Microscopic studies with pharmacological maneuvers are required to address these questions.

Finally, although the majority of the blood within the myocardium is resident in the capillaries, a fraction (10%) is present in small arterioles and venules <200 µm in diameter (16). Changes in dimensions of these vessels could, therefore, produce changes in myocardial VI on MCE. In the presence of a stenosis that can limit either resting or hyperemic flow, however, these vessels should be maximally dilated and should therefore increase myocardial VI. Since we noted a decrease in myocardial VI in these settings, we believe that our results pertain to capBV.

In conclusion, our study breaks new ground in the understanding of the control of CBF. It demonstrates the vital role played by capillaries in the control of CBF during hyperemia, both in the absence and presence of noncritical stenosis. It also suggests that capillaries may be responsible for the decrease in resting CBF caused by severe stenosis. Our results imply that alterations in capillary resistance may explain the abnormal CBF reserve seen in conditions not associated with coronary stenosis. They also offer an alternate explanation for the critical coronary closing pressure. These results have far-reaching implications not only in our understanding of coronary physiology but also in the use of myocardial perfusion imaging.