Disentangling the Gordian knot of local metabolic control of coronary blood flow

Abstract

Recognition that coronary blood flow is tightly coupled with myocardial metabolism has been appreciated for well over half a century. However, exactly how coronary microvascular resistance is tightly coupled with myocardial oxygen consumption (MV̇o₂) remains one of the most highly contested mysteries of the coronary circulation to this day. Understanding the mechanisms responsible for local metabolic control of coronary blood flow has been confounded by continued debate regarding both anticipated experimental outcomes and data interpretation. For a number of years, coronary venous Po₂ has been generally accepted as a measure of myocardial tissue oxygenation and thus the classically proposed error signal for the generation of vasodilator metabolites in the heart. However, interpretation of changes in coronary venous Po₂ relative to MV̇o₂ are quite nuanced, inherently circular in nature, and subject to confounding influences that remain largely unaccounted for. The purpose of this review is to highlight difficulties in interpreting the complex interrelationship between key coronary outcome variables and the arguments that emerge from prior studies performed during exercise, hemodilution, hypoxemia, and alterations in perfusion pressure. Furthermore, potential paths forward are proposed to help to facilitate further dialogue and study to ultimately unravel what has become the Gordian knot of the coronary circulation.

INTRODUCTION

Mythology provides the legend of Gordius, who upon becoming king of Phrygia, dedicated his chariot to Zeus and fastened it to a pole with an impossible knot. An oracle predicted that whoever could disentangle the knot would be the future king of Asia. Many tried and failed, until an impatient Alexander the Great took out his sword and solved the conundrum by merely slicing through the knot. The “Gordian knot” has thus come to symbolize a complex and impossible problem to solve. This metaphor is applicable to the coronary circulation where coronary blood flow is inextricably tied to myocardial metabolism as perfusion not only dictates the substrate supply for, but remains primarily dependent on, the level of oxidative phosphorylation (9, 27, 47). The fundamental coupling between coronary flow and metabolism is essential in a highly metabolically active tissue such as the heart, where baseline levels of oxygen extraction typically exceed 70% (28, 40, 105). It is within these basal constraints that local metabolic mechanisms are proposed to predominate to ensure that alterations in myocardial oxygen consumption (MV̇o₂) are balanced by commensurate changes in myocardial oxygen delivery (Fig. 1A) such that cardiac contractile function and output are adequately maintained across a wide range of (patho-)physiologic perturbations (106). Exactly how coronary microvascular resistance is precisely coupled with the underlying level of MV̇o₂ remains one of the most highly contested mysteries of the coronary circulation to this day.

Fig. 1.

Classic relationships between coronary blood flow and myocardial oxygen consumption (A), coronary blood flow and coronary venous Po₂ (B), and coronary venous Po₂ and myocardial oxygen consumption (C) in dogs.

The central tenet of the local metabolic hypothesis of coronary blood flow control proposes that production of vasodilator metabolites corresponds with reductions in myocardial tissue Po₂ and thus acts in a negative feedback manner to preserve tissue oxygenation within normal physiological limits (23, 27, 40, 47). Accordingly, the ability to assess how myocardial tissue Po₂ is influenced by variations in key determinants of myocardial perfusion (i.e., MV̇o₂, arterial oxygen levels and coronary perfusion pressure) is essential for the delineation of mechanisms responsible for local metabolic control of coronary blood flow. To this end, coronary venous Po₂ has been proposed, and generally accepted, to directly represent underlying changes in myocardial oxygenation (23, 40, 58). This theory is directly supported by data that demonstrate that coronary blood flow is progressively increased as coronary venous Po₂ (i.e., myocardial tissue Po₂) falls below a critical threshold value of ~18–20 mmHg in conscious dogs at rest and during exercise induced increases in MV̇o₂ (Fig. 1B) (106). As such, changes in coronary venous Po₂ relative to MV̇o₂ have been utilized as a means to directly assess the balance between coronary blood flow and MV̇o₂ (Fig. 1C) (27, 28, 105, 106). Within this paradigm, if increases in coronary blood flow perfectly match increases in MV̇o₂, coronary venous Po₂ remains unchanged as cardiac workload is elevated. Alternatively, if/when a pathway that contributes to local metabolic coronary vasodilation is inhibited, the impaired flow response results in reductions in myocardial oxygenation and enhanced oxygen extraction, both of which act to significantly steepen the relationship between coronary venous Po₂ and MV̇o₂ (27, 47).

Herein begins the circular reasoning that is inherent to the coupling between coronary blood flow and myocardial metabolism (Fig. 2). First, it is undisputed that MV̇o₂ is the primary determinant of coronary blood flow. However, it must be recognized that inhibition of any mechanism that contributes to metabolism-mediated increases in coronary flow will diminish myocardial oxygen delivery. Furthermore, if/when this reduction is severe enough, MV̇o₂ (the principal factor driving the flow response) will be decreased commensurate with the degree of fall in oxygen delivery (47, 105). Second, coronary venous Po₂ simultaneously represents the proposed “error signal” responsible for the production of local metabolites (i.e., myocardial tissue Po₂) as well as the resultant of the overall balance between myocardial oxygen supply and consumption (106). While this is a logical extension of any variable controlled by a negative feedback system, conclusions regarding changes (or lack thereof) in coronary venous Po₂ can prove quite challenging given that a perfect balance between coronary flow and MV̇o₂ (Fig. 1C) is not consistent with reductions in tissue Po₂ driving the response (Fig. 1B). Lastly, interpretation of changes in coronary venous Po₂ is also confounded by the fact that increases in myocardial oxygen extraction reduce coronary venous Po₂, thus leaving one to question whether alterations in coronary venous Po₂ are driven primarily by changes in myocardial oxygen extraction, myocardial tissue Po₂, or perhaps both. Taken together, the dynamic interplay between these essential elements of coronary control continues to generate much debate regarding the most appropriate means by which to assess the balance between coronary blood flow and MV̇o₂. Below we consider several situations that highlight the difficulties in interpreting this complex interrelation and the arguments that emerge from the different conclusions that can be drawn from these scenarios. Finally, we consider potential paths forward to help resolve this physiologic Gordian knot by proposing updated criteria of local metabolic mechanisms that seek to simplify analyses and downplay constraints of continued overreliance on interpretation of changes in coronary venous Po₂. This review is principally focused on short-term adaptations in coronary blood flow (i.e., acute intervention studies) but not on chronic changes to exercise, pressure, or flow and does not aim to provide a detailed discussion of individual metabolites that have been proposed to contribute to metabolic coronary vasodilation. For these aspects, readers are referred to several comprehensive reviews by Duncker et al. (27, 30), Canty et al. (14), and Goodwill et al. (47).

Fig. 2.

Schematic representation of the interrelations between the primary variables proposed to modulate local metabolic control of coronary blood flow.

DIFFICULTIES WITH ASSESSMENT AND INTERPRETATION OF LOCAL METABOLIC CONTROL

There is a substantial amount of evidence that the relationship between coronary blood flow and MV̇o₂ is linear and extremely consistent across many species, including mice (84), rats (41), dogs (3), swine (70), horses (77), and humans (53) (Fig. 3; heart rate used as index of MV̇o₂) [see Duncker et al. (27–29) for more detailed review]. However, it should be appreciated that reported normalized values of coronary blood flow (per gram myocardium perfused) in mice are substantially (up to an order of magnitude) higher compared with other (larger) species (84). Nonetheless, within this highly conserved consistency it is reasonable to surmise that disruption of any pathway that contributes to balance between oxygen delivery and metabolism should produce proportional reductions in the slope of this relationship. However, practical application of this premise is flawed for a few reasons. Initially, it is important to recognize that MV̇o₂ itself is typically not measured but calculated from the Fick principle, i.e., coronary blood flow times myocardial oxygen extraction (arterial minus coronary venous oxygen content) (47). As such, plotting coronary blood flow (or coronary venous Po₂) relative to MV̇o₂ results in a certain degree of redundancy and raises questions of validity as specific variables are plotted relative to themselves (57). Next, the relationship between coronary blood flow and MV̇o₂ operates near the maximum level of oxygen extraction, and thus the slope of this relationship cannot be appreciably decreased within these tight physiologic constraints. Consequently, alterations in the slope of the coronary blood flow versus MV̇o₂ relationship cannot be utilized as a means to reliably assess the contribution of specific mechanisms of local metabolic coronary vasodilation [see Goodwill et al. (47) for more in depth review]. Thus the field has largely relied on assessment of changes in coronary venous Po₂ relative to MV̇o₂ as a means to interpret the overall balance between myocardial oxygen delivery and metabolism (27, 47, 58, 105, 106).

Fig. 3.

Relationship between coronary blood flow and heart rate [index of myocardial oxygen consumption (MV̇o₂)] across multiple species: mouse (84), rat (41), dog (3), swine (70), horse (77), and human (53).

Interpretation of Cause Versus Effect Across Conditions and Species

Classically, studies in conscious, instrumented dogs reliably demonstrate significant reductions in coronary venous Po₂ with exercise-induced increases in MV̇o₂ (Fig. 4) (27, 47, 49, 106). These data, which represent global, timed averaged snapshots of these variables, support the hypothesis that reductions in myocardial tissue Po₂ (oxygen-sensing mechanisms) are responsible for, or at least significantly contribute to, the production of local metabolites and increases in coronary blood flow (Fig. 4A) (52, 111). However, it is evident that these reductions in coronary venous Po₂ in dogs (typically from ~20 mmHg at rest to ~15 mmHg with exercise) are not linear and characteristically plateau as MV̇o₂ increases above ~150 μl O₂·min·⁻¹·g⁻¹ (Fig. 4B). This response lies in stark contrast with data from swine (43, 54, 81) and humans (53, 55), which demonstrate relatively modest changes (~0–1 mmHg) in coronary venous Po₂ with approximately two- to fourfold increases in MV̇o₂; i.e., relationship between coronary venous Po₂ and MV̇o₂ remains essentially flat (Fig. 4B). One reason for these discrepant results is the involvement of different mechanisms/pathways that modulate coronary responses to increased MV̇o₂ in a species-dependent manner. In particular, prior studies of autonomic control of coronary blood flow in dogs determined that inhibition of α-adrenergic receptors significantly reduces the slope of the relationship between coronary venous Po₂ and MV̇o₂ (27, 47, 52). In contrast, α-adrenergic blockade has no effect on coronary venous Po₂ at rest or during increases in MV̇o₂ in swine (31, 96). Thus paradoxical α-adrenergic vasoconstriction appears to largely account for the distinct relationship between coronary venous Po₂ and MV̇o₂ reported in dogs (Fig. 4B). Taken together, these findings collectively argue against a requisite role for myocardial tissue Po₂ as the error signal for local metabolic vasodilation.

Fig. 4.

Relationship between coronary blood flow and coronary venous Po₂ (A) and coronary venous Po₂ and myocardial oxygen consumption (B) across multiple species: dog (38, 51, 110), swine (43, 54, 81), and human (53, 55).

The discrepancies outlined above also relate to the more recent proposal of the adenine-nucleotide hypothesis (38, 51). This relatively new paradigm postulates that erythrocytes act as oxygen sensors in coronary capillary beds, wherein hemoglobin saturation declines and results in the release of ATP with subsequent activation of endothelial purinergic receptors and the initiation/propagation of a retrograde conducted vasodilator response (49). While there is evidence to support this hypothesis (38, 51, 87), the effective relationship between coronary blood flow and coronary venous hemoglobin saturation (Fig. 5) produces qualitatively similar responses to that of coronary venous Po₂ across species (Fig. 4). Here again, data from exercising dog studies are most consistent with this postulate as reductions in coronary venous hemoglobin saturation of up to 10–12% are associated with an ~65% increase in coronary venous plasma ATP concentration and a two- to threefold increase in coronary blood flow (38, 51). In contrast, markedly higher slopes are observed in swine (43, 81) and humans (36, 55), with 1–5% reductions in coronary venous hemoglobin saturation corresponding with two- to fourfold increases in coronary blood flow (Fig. 5). Given these findings, and the fact that the unique relationship in dogs is essentially abolished by inhibition of α-adrenergic receptors (27, 47, 52), it seems unlikely that such modest changes in hemoglobin saturation are the primary sensor responsible for linking myocardial oxygen delivery with metabolism. Furthermore, questions and inconsistencies regarding interpretation of changes in coronary venous Po₂ are also relevant to reported alterations in coronary venous hemoglobin saturation by virtue of their essential physiologic relationship.

Fig. 5.

Relationship between coronary blood flow and coronary venous hemoglobin saturation across multiple species: dog (38, 51), swine (43, 81), and human (36, 55).

On the surface, reasons to question whether measurements of coronary venous Po₂ reflect a given level of myocardial tissue oxygenation are few. Although evidence of arteriovenous shunting and countercurrent exchange in the heart has been documented (6, 16, 24, 92), diffusional shunting of oxygen is quite limited and impeded by hemoglobin binding (63), i.e., consistent with very low values of coronary venous Po₂ reported across multiple species (27, 40, 47). However, it is important to appreciate that species differences in coronary vascular structural patterns exist (5, 42, 44, 60, 62) and that reported values of coronary venous Po₂ (and other primary response variables) represent global steady-state values that may not necessarily reflect the highly dynamic processes that correspond with oscillatory (20, 94, 112, 119) and transmurally heterogeneous variations (27, 104, 105) across the coronary circulation. These phenomena lead to issues with variability in coronary response data both within and between studies, which are further complicated by inconsistencies in the measurements themselves. While recognition of these confounding influences is critical, evaluation of static measures of key coronary response variables clearly provides insight into the regulation of coronary blood flow in response to a variety of patho-physiologic perturbations. Nevertheless, interpretation of changes in these variables, such as coronary venous Po₂, can be quite nuanced. This point is illustrated by comparing coronary responses to anemia versus hypoxemia. Numerous studies have demonstrated that increases in coronary blood flow to reductions in hematocrit (hemodilution) (18, 19, 46, 65, 76, 113) or arterial Po₂ (hypoxemia) (37, 45, 56, 75, 82, 114, 115) are quite similar (Fig. 6A) and that these increases in flow are sufficient to maintain overall levels of myocardial oxygen delivery (Fig. 6B). Despite these comparable effects, which reportedly occur over relatively similar ranges of MV̇o₂ (~50–125 μl O₂·min·⁻¹·g⁻¹), coronary venous Po₂ consistently falls in response to hypoxemia (Y = 0.94x + 6.4; r = 0.75) yet remains unchanged in response to hemodilution (Y = 0.14x + 20.5; r = 0.28; Fig. 6C; P value for slope < 0.01). Such divergent responses have been demonstrated across a wide variety of species, including dogs (18, 19, 37, 56, 76, 82, 114), lambs (74), swine (45, 65, 68, 113), and humans (46). Based on the apparent preservation of oxygen delivery at comparable levels of MV̇o₂ in both cases, it seems unlikely that the reduction of coronary venous Po₂ in response to hypoxemia represents a progressive decline of myocardial tissue oxygenation. While these divergent responses implicate the presence of distinct vasodilator mechanisms, underlying differences between viscosity, shifts in oxy-hemoglobin dissociation curve, and/or diffusion properties must also be acknowledged (76, 116). Interestingly, data from the Feigl laboratory (7, 95) demonstrate discrepant, nonlinear relationships between myocardial oxygenation (assessed by mean myoglobin saturation) and venous Po₂ in isolated, buffer-perfused guinea pig hearts in the absence and presence of erythrocytes (5% hematocrit). Thus, while it is readily apparent that the heart possesses exquisitely sensitive oxygen-sensing mechanisms that act to maintain adequate myocardial oxygen delivery, exactly how these pathways are sensed and regulated remains poorly understood [see Duncker et al. (27) and Goodwill et al. (47) for more comprehensive review].

Fig. 6.

Relationship between coronary blood flow (A), myocardial oxygen delivery (B), coronary venous Po₂ (C) relative to arterial oxygen content in response to hemodilution (solid line; Refs. 18, 19, 46, 65, 76, 113) and hypoxemia (dashed line; Refs. 37, 45, 56, 74, 82, 114, 115). Arterial oxygen content was used to normalize the level of oxygen deficit in each study.

Local Metabolic Control in Response to Changes in Coronary Perfusion Pressure

The coronary circulation has the intrinsic ability to maintain blood flow relatively constant over a wide range of perfusion pressures (1). This autoregulatory capacity is especially crucial in times of coronary stenosis, where hypoperfusion can lead to rapid reductions in cardiac function and myocardial injury (39, 59). While this phenomenon is distinctly different from a pure metabolic stimulus (e.g., exercise) in that it involves changes in perfusion pressure and myocardial ischemia, the most prominent and accepted theory to explain coronary pressure-flow autoregulation centers around local metabolic regulation of coronary vascular resistance (23, 39). In this context, the local metabolic hypothesis proposes that myocardial oxygen tension determines the coronary autoregulatory capacity by increasing the production of vasodilator metabolites as perfusion pressure is reduced. This contention is supported by the classic study of Dole and Nuno (23), which demonstrated that coronary venous Po₂ decreases with coronary perfusion pressure and that autoregulatory gain [equal to

$$1-\frac{\Delta \text{Flow}/\text{Flow}}{\Delta \text{Pressure}/\text{Pressure}}$$

where a value of 1 represents perfect autoregulation] is dependent on normal physiologic levels of myocardial tissue Po₂, i.e., coronary venous Po₂ < 32 mmHg. This association between coronary venous Po₂ and perfusion pressure has been demonstrated by a number of laboratories in dogs (4, 100) and swine (11, 64). Thus there is substantial evidence to support that mechanisms closely associated with reductions in coronary venous Po₂ are directly associated with decreases in coronary microvascular resistance in response to changes in perfusion pressure and/or hypoxemia (25). However, whether the reduction in coronary venous Po₂ reflects a cause of autoregulatory behavior or a consequence of lowered perfusion pressure and augmented myocardial oxygen extraction remains unclear.

To examine the extent to which coronary autoregulatory behavior is contingent on normal physiologic levels of coronary venous Po₂, Kiel et al. (64) examined coronary responses to changes in perfusion pressure (140–40 mmHg) in the absence and presence of euvolemic anemia (50% reduction in hematocrit) and infusion of dobutamine to augment MV̇o₂. Hemodilution was utilized in this study as reductions in hematocrit diminish coronary vasomotor tone with little/no change in coronary venous Po₂ (17, 65, 76). Consistent with prior studies supporting local metabolic control mechanism in the autoregulatory response (4, 23), autoregulatory gain (determined by changes in coronary flow in response to 20 mmHg increments at pressures ranging from 120 to 60 mmHg) remains relatively constant over coronary venous Po₂ values ranging from ~30 to ~10 mmHg (Fig. 7B; Y = −0.01x + 0.6; r = 0.24). In contrast, autoregulatory capacity significantly decreased (P value for slope = 0.03) over a similar range of coronary venous Po₂ values in the presence of hemodilution and dobutamine (Fig. 7B; Y = 0.03x −0.7; r = 0.55). These findings indicate that the local metabolic hypothesis is not sufficient to explain autoregulatory behavior. Interestingly, additional data from the study of Kiel et al. (64) suggest that the primary mechanism of autoregulation could be more myogenic in origin, as coronary zero flow pressure (index of overall vascular tone) was highly predictive of changes in flow and autoregulatory gain. Nevertheless, this conclusion, along with those for other scenarios such as exercise, hemodilution, and hypoxemia, rests on the general assumption that has been applied to coronary studies for decades; i.e., coronary venous Po₂ is a valid measure of myocardial Po₂. However, each of these examples highlight clear inconsistencies that not only obscure interpretation but provide reasons to question whether this is truly always the case or not (21).

Fig. 7.

Relationship between coronary blood flow and coronary perfusion pressure (A) and autoregulatory gain (change in coronary flow over 20-mmHg increments at pressures ranging from 120 to 60 mmHg) relative to coronary venous Po₂ (B). Responses are plotted in the absence (solid line) and presence of euvoluemic hemodilution (dashed line) (~50% reduction in hematocrit) plus dobutamine (increase heart rate ~75–100% above baseline levels). Data from Kiel et al. (64).

Additional Considerations in Interpreting Potential Local Metabolic Mechanisms

Traditional interpretation of coronary response variables has held that inhibition of a factor that has a sustained influence on coronary microvascular resistance will produce a reduction in coronary venous Po₂ (or venous saturation) across a comparable range of MV̇o₂, with little/no effect on coronary blood flow (27, 28, 47, 49, 51, 105, 107). One classic example of this are studies that have demonstrated a parallel downward shift in the relationship between coronary venous Po₂ and MV̇o₂ following the inhibition of nitric oxide synthase (Fig. 8B). These studies have largely concluded that nitric oxide exerts a modest, tonic vasodilator influence in the coronary circulation but that it is not required for local metabolic coronary vasodilation (2, 26, 61, 80, 98, 109). However, this downward shift is associated with a relatively modest increase in the relationship between coronary blood flow and MV̇o₂ (Fig. 8A). One hypothesis to explain these divergent responses is that reductions in coronary venous Po₂ following nitric oxide synthesis inhibition reflect an augmented error signal for metabolic vasodilation such that the magnitude of exercise-mediated coronary vasodilation is slightly augmented (51). Another hypothesis that has not been readily considered is that nitric oxide itself has been demonstrated to be an enhancer of oxygen transfer from erythrocytes, i.e., produces a right shift of the oxygen-hemoglobin dissociation curve (66, 67). Consequently, inhibition of nitric oxide production would be postulated to result in a leftward shift of the dissociation curve, thereby providing a potential explanation for a reduction in coronary venous Po₂ in the absence of any reduction in coronary blood flow or MV̇o₂. Another example of such an effect involves the disruption of adenosine receptor signaling, which has been proposed as a target of sickle cell disease as adenosine augments 2,3-diphophoglycerate (2,3 DPG) production from erythrocytes (120). Thus nonselective inhibition of adenosine receptors with antagonists such as 8-phenyltheophylline (8-PT) would also be predicted to produce a leftward shift of the oxyhemoglobin dissociation curve. Such a consequence could indeed explain the paradoxical findings that 8-PT has no effect on coronary blood flow yet produces a parallel downward shift of the relationship between coronary venous Po₂ and MV̇o₂ (32, 61, 73, 110) at the same time as measured values of adenosine are reported to be below the threshold value necessary for vasodilation to occur (110). Accordingly, more rigorous consideration of potential influences of antagonists on hemoglobin binding affinity is warranted before definitive conclusions regarding the role of specific pathways can be made.

Fig. 8.

Relationship between coronary blood flow (A) and coronary venous Po₂ (B) vs. myocardial oxygen consumption in conscious instrumented dogs at rest and during graded treadmill exercise. Responses are plotted in the absence and presence of nitric oxide synthesis inhibition with N-nitro-l-arginine. Data replotted from Tune et al. (109).

Data that chronic alterations in oxygen transport in the coronary circulation can influence coronary response variables have also been documented. In particular, prior studies in exercise trained animals suggest that augmented levels of myocardial oxygen extraction, hence lower values of coronary venous Po₂ (117), are likely related to training-induced increases in the capillary transport capacity (69, 71, 72, 85, 118). Conversely, reductions in coronary venous Po₂ in obese (10, 13, 97, 103) and diabetic, dyslipidemic (8, 99) swine have been shown to be directly associated with reductions in capillary density (rarefaction) and overall reductions in myocardial lactate consumption (13). The presence of capillary rarefaction in chronic disease would act to reduce oxygen extraction capacity such that reductions in coronary venous Po₂ likely reflect impaired myocardial oxygenation. Regardless, potential alterations in the coronary circulation should be duly considered when interpreting changes in coronary venous Po₂ in the setting of chronic disease states.

APPROACHES TO ASSESS LOCAL METABOLIC CONTROL OF CORONARY BLOOD FLOW

The ability to resolve what has been the inextricable Gordian knot of the coronary circulation lies first in the general agreement of the most appropriate means by which to assess the question. Unfortunately, this essential component remains contested as members of the field continue to debate the paradox of whether coronary venous Po₂ (Fig. 4) or hemoglobin saturation (Fig. 5) represents the error signal (cause) for metabolic vasodilation (51, 87) versus the consequence of an imposed imbalance between flow and metabolism (27, 28, 47, 106) or somehow both the beginning and end of the knot itself. Accordingly, it has been essentially impossible to arrive at any consensus when differences in anticipated experimental outcomes and data interpretation persist. Below we propose a potential, more straightforward path for assessing and interpreting the contribution of local metabolic mechanisms based on essential underlying experimental findings.

Fundamental a Priori Prediction

When considering how to determine whether a specific factor or pathway contributes to local metabolic coronary vasodilation, the most fundamental assumption that can be made, a priori, is that inhibition of the factor (or pathway) should diminish metabolic-mediated increases in coronary blood flow. As outlined above, such an effect cannot manifest as a marked reduction in the slope of the relationship between coronary blood flow and MV̇o₂, but there is evidence that this response results in a commensurate decrease in MV̇o₂ itself. There are two primary examples of this in the literature. First, studies in both dogs (93) and swine (10, 48) have demonstrated that inhibition of voltage-gated K⁺ (K_V) channels diminishes coronary blood flow (Fig. 9A) and MV̇o₂ (Fig. 9B), during cardiac pacing, catecholamine infusion, and exercise-mediated increases in myocardial metabolism. Interestingly, despite baseline reductions in coronary venous Po₂, these changes were not accompanied by progressive reductions in venous Po₂ as cardiac workload was elevated (Fig. 9C). Alternatively, studies in mice lacking smooth muscle K_V1.5 channels demonstrated marked reductions in the relationship between myocardial blood flow and cardiac work, diminished myocardial tissue Po₂, and cardiac contractile dysfunction (84). There are several potential explanations for the more modest effect of K_V channel blockade in large animals, including partial antagonism with pharmacologic inhibition, steady-state measurements in the large animal studies versus dynamic state measurements in the mouse studies, and/or that myocardial Po₂ and coronary venous Po₂ are distinctly different variables. Regardless, the truncation of the MV̇o₂ response adheres to the law of mass balance and fits with the primary expectation of antagonist effect; i.e., reduction of the overall flow response limits the ability to increase MV̇o₂. For these reasons, pathways that converge on K_V channels are proposed to be involved in local metabolic control of coronary blood flow (47, 105). However, corresponding reductions in coronary flow and MV̇o₂ elicit a “chicken or egg” argument as to whether the response is mediated by increases in coronary microvascular resistance versus primary reductions in myocardial metabolism.

Fig. 9.

Coronary blood flow (A), myocardial oxygen consumption (B), and coronary venous Po₂ (C) at rest and during exercise in the absence and presence of the K_V channel inhibitor 4-aminopyridine (4-AP) in conscious instrumented swine. Data from Berwick et al. (10). *P < 0.05 vs. control.

Another case consistent with this axiom are studies that have examined redundant and compensatory mechanisms invoked in response to the inhibition of ATP-sensitive K⁺ (K_ATP) channels. These studies reliably show that inhibition of K_ATP channels with glibenclamide produces reductions in coronary blood flow and MV̇o₂ at rest and during increases in metabolism in dogs (33, 88, 89) as well as swine (78, 79). Furthermore, studies by the Bache laboratory (34, 35, 61) demonstrate that the addition of adenosine receptor (8-PT) and nitric oxide synthesis [N-nitro-l-arginine (LNNA)] inhibitors results in further progressive reductions in the coronary blood flow response to graded treadmill exercise (Fig. 10A). These reductions in coronary flow were accompanied by decreases in cardiac contractile function (systolic wall thickening) and MV̇o₂ (Fig. 10B) (34, 35, 61), as would be expected with impairments in myocardial oxygen delivery. These data indicate reductions in coronary flow produced by glibenclamide were sufficient to induce compensatory increases in myocardial adenosine release and in the contribution of nitric oxide to exercise-induced increases in coronary blood flow. Importantly, subsequent experiments by the Bache laboratory (35, 61) directly addressed the chicken or egg conundrum by demonstrating that the restoration of coronary blood flow to normal, untreated control values (with sodium nitroprusside, which by itself had no effect on regional contractile function) in the presence of glibenclamide and 8-PT (35) completely restored values of MV̇o₂ (Fig. 10C) and systolic wall thickening to normal levels under baseline resting conditions (Fig. 10D). These data are critical in confirming that the reductions in MV̇o₂ in the presence of glibenclamide and 8PT were related to deficits in coronary blood flow and not to a direct negative inotropic effect or to inhibition of mitochondrial respiration. These data are further corroborated by the lack of effect of the mitochondrial selective K_ATP channel blocker 5-hydroxydecanoate (15). Taken together, these findings provide direct evidence that inhibition of a specific pathway that contributes to the regulation of coronary blood flow diminishes the coronary flow response and is accompanied by proportional (and correctable) reductions in MV̇o₂. Further support for this contention can be found in prior studies designed to test the adenosine hypothesis, in that inhibition of adenosine signaling was shown to diminish exercise-induced increases in coronary blood flow only in the presence of a coronary stenosis (73). In other words, reductions in coronary blood flow are indeed evident when a specific factor involved in the response is inhibited.

Fig. 10.

Coronary blood flow (A) and myocardial oxygen consumption (B) at rest and during exercise (Ex) in the absence and presence of the K_ATP channel inhibitor glibenclamide + the adenosine receptor blocker 8-phenytheophylline (8-PT) ± the nitric oxide synthase inhibitor [N-nitro-l-arginine (LNNA)] in conscious instrumented dogs. Myocardial oxygen consumption (C) and systolic wall thickening (D) vs. coronary blood flow with and without glibenclamide (Glib) + 8-PT and the vasodilator sodium nitroprusside (SNP) under baseline resting conditions [data from Duncker et al. (35) and Ishibashi et al. (61)]. *P < 0.05 vs. control. †P < 0.05 vs. glibenclamide + 8PT.

It is also important to consider the extent to which the attenuation of the coronary flow and MV̇o₂ response influences coronary venous Po₂. Specifically, while inhibition of either K_V or K_ATP channels alone produces reductions in resting coronary venous Po₂, the classically predicted steepening (clockwise rotation shown in Fig. 1C) of the relationship between coronary venous Po₂ and MV̇o₂ is not readily apparent in either dog (34, 35, 93) or swine (10, 48) studies (Fig. 11). We submit that the lessening of coronary venous Po₂ (truncated, parallel shift) reflects the consequence of antagonist-mediated decreases in myocardial oxygen delivery-metabolism balance. However, cases of the steepening of the relationship between coronary venous Po₂ and MV̇o₂ have been documented in the presence of combined K_ATP channel (high-dose, intracoronary glibenclamide), adenosine receptor, and nitric oxide synthesis inhibition (61) as well as in chronic disease conditions such as obesity-metabolic syndrome (10). Such progressive increases in myocardial oxygen extraction (diminished delivery-metabolism balance) appear to only occur under conditions in which coronary blood flow is substantially diminished, typically ≥ 40% at higher levels of exercise. As proposed above, findings of truncated, parallel shifts versus steepened slopes are likely dependent on the degree of pharmacologic antagonism and/or the level of physiologic compensation. Regardless, we propose that antagonist-induced reductions in coronary blood flow and MV̇o₂ (correctable with vasodilator infusion), along with accompanied truncation and reductions in indexes of the balance between coronary blood flow and myocardial metabolism, are the most fundamental experimental findings that must be observed to satisfy an obligatory role for any factor or pathway in local metabolic coronary vasodilation.

Fig. 11.

Coronary venous Po₂ vs. myocardial oxygen consumption at rest and during exercise in the absence and presence of the Kv channel blocker 4-aminopyridine [4-AP; data from Berwick et al. (10); A] and the K_ATP channel inhibitor glibenclamide [data from Duncker et al. (35); B].

Updated Criteria for Defining Local Metabolic Mechanisms

In 1983, Feigl proposed criteria for the classification of specific factors as local “metabolites” (40). These standards, a modification of Koch’s classic postulates, include the necessary biochemical machinery for the production and release of the metabolite to be present, along with appropriate mechanisms for inactivation and/or reuptake. Further experimental requirements are that the metabolite is quantifiably released under the appropriate conditions, that administration of the metabolite mimics the desired physiologic response, and that inhibition of the metabolite has effects consistent with the hypothesis. Without a doubt, differences in opinion of what finding(s) constitute the predicted inhibitory effects have significantly clouded interpretation and conclusions regarding mechanisms of local metabolic control to this point. To this end, we propose the following updates to the criteria for defining mechanisms of local metabolic control:

• 1. Quantitative studies demonstrate that increases in MV̇o₂ correspond with increases in metabolite concentration (release). Such studies include direct measures via placement of myocardial electrodes, microdialysis probes, and/or collection of coronary venous blood. However, limitations of these approaches including myocardial damage and the potential for rapidly degraded substances to not appear in venous blood should be recognized.
• 2. Physiologically relevant levels of the released metabolite are sufficient to produce coronary vasodilation in a concentration-dependent manner. Experiments are needed to validate that physiologic concentrations of the metabolite are sufficient to produce coronary vasodilation. Recognition that intracoronary administration of the metabolite (or related agonist) does not specifically equate with myocardial-dependent release and responses should be appreciated.
• 3. Inhibition of metabolite production and/or receptor signaling diminishes coronary blood flow at rest and/or in response to increases in MV̇o₂. A requisite role of a metabolite in the control of coronary microvascular resistance is demonstrated by dose-dependent reductions in coronary blood flow in response to metabolite blockade. Furthermore, if the effect of the metabolite is only present at rest, inhibition will diminish baseline coronary blood flow but not the overall increase (delta) of coronary response to increases in MV̇o₂. Alternatively, if the metabolite contributes only during increases in MV̇o₂, then inhibition will not influence baseline flow but will diminish MV̇o₂-mediated increases in flow. If the metabolite involvement is at rest and during increases in MV̇o₂, then inhibition will decrease coronary flow and rest and in response to augmented MV̇o₂.
• 4. Reductions in coronary blood flow produced by inhibition of the metabolite correspond with commensurate decreases in MV̇o₂ (and indexes of cardiac function) and indexes of the balance between coronary blood flow and metabolism. Findings with regard to effects of metabolite inhibition on coronary blood flow demonstrate proportionate reductions in MV̇o₂ and indexes of cardiac function. Collectively, the data should not violate mass balance and inhibition results in the truncation of the relationships of coronary blood flow and coronary venous Po₂ relative to MV̇o₂.
• 5. Corresponding reductions of MV̇o₂ (and indexes of cardiac function) are reversed by vasodilator-mediated restoration of coronary blood flow to normal (untreated) levels. Studies to demonstrate that reductions in MV̇o₂ and cardiac function produced by metabolite inhibition are related to deficits in myocardial oxygen delivery and not to a direct negative inotropic effect or to an impairment in mitochondrial respiration, are imperative.

Despite well over 50 years of dedicated research in this area, no study has documented evidence to support that any specific metabolite fulfills these criteria. Furthermore, there are only a couple factors, most notably H₂O₂ (90, 91, 93) and ATP (adenine nucleotides) (38, 50, 51, 87), that satisfy some of these criteria and merit further consideration. Interestingly, these metabolites have been shown to converge on K⁺ channels (22); however, definitive data to establish causation are presently lacking [see Duncker et al. (27) and Goodwill et al. (47) for more comprehensive review]. Another mechanism that should be recognized is that of the dual role of catecholamines, which contribute to both arms of the delivery/metabolism equation in mediating β-adrenergic vasodilation and increases in heart rate and contractility (27, 47). This pathway is proposed to occur without an error signal (feedforward or open-loop control) and thus could explain how coronary venous Po₂ is maintained over a wide range of MV̇o₂ (Fig. 4). However, precisely how autonomic β-adrenergic pathways work with proposed parallel oxygen-sensing pathways remains to be determined. Importantly, the criteria outlined above are appropriate for both feedback and feedforward pathways. Additionally, while the presence of alternative parallel and/or redundant pathways that can act in concert or in a compensatory manner must be appreciated (34, 35, 61, 78, 108), there is ample evidence to support the assumption of a (correctable) truncation of the coronary flow/MV̇o₂ response when a specific pathway that contributes to the regulation of coronary blood flow is inhibited (10, 35, 61, 93) (see Fig. 10). We propose that the application of the updated criteria above, which appear particularly suited for short-term studies using acute interventions, will help to facilitate interpretation and understanding of mechanisms responsible for metabolic control of coronary blood flow.

SUMMARY AND IMPLICATIONS

The local metabolic hypothesis of blood flow control remains one of the primary questions of coronary physiologists to this day. However, discerning the mechanisms responsible for equilibrating alterations in coronary microvascular resistance with underlying changes in MV̇o₂ has been confounded by the continued debate around both anticipated experimental outcomes and data interpretation. For a number of years, coronary venous Po₂ has been generally accepted to represent a measure of myocardial tissue oxygenation (12, 27, 28, 40, 47, 58, 105, 106) and thus the classically proposed error signal for the generation of vasodilator metabolites in the heart. Yet, evidence from the coronary literature across species indicates that reductions in coronary venous Po₂ (tissue oxygenation) are not required for coupling coronary flow with MV̇o₂ in response to exercise (Fig. 4), reductions in arterial oxygenation (Fig. 6), or changes in perfusion pressure (Fig. 7). Similar questions regarding hemoglobin saturation serving as the primary sensor for linking myocardial oxygen delivery with metabolism also remain. Together these findings importantly argue against the original formulation of the metabolic hypothesis and implicate the presence and involvement of alternative oxygen- or metabolic (e.g., ADP/ATP ratio)-sensing mechanisms within the vasculature and/or myocardium that are capable of maintaining myocardial oxygen delivery/metabolism balance in response to a variety of (patho-)physiologic perturbations. Nevertheless, interpretation of changes in coronary venous Po₂ (or coronary venous hemoglobin saturation) relative to MV̇o₂ is quite nuanced, inherently circular in nature, and subject to the largely unaccounted for potential for confounding alterations in oxy-hemoglobin dissociation and/or myocardial diffusion properties. Accordingly, a more straightforward path for assessing and interpreting the contribution of local metabolic mechanisms is greatly needed to begin to better understand this fundamental physiologic phenomenon.

We propose that the simplest a priori prediction that can be made is that inhibition of a factor or pathway that contributes to local metabolic control should attenuate the coronary flow response relative to the untreated control response. Prior studies demonstrate that this is indeed the case to the extent that imposed reductions in myocardial oxygen delivery result in commensurate, yet correctable decreases in MV̇o₂ (Figs. 9 and 10). Such effects are importantly accompanied by anticipated reductions in coronary venous Po₂ (Fig. 11) and are evident in scenarios that involve activation of compensatory vasodilator pathways (Fig. 10). We submit that decreases in coronary venous Po₂ (or coronary venous hemoglobin saturation) in the absence of clear reductions in coronary blood flow and MV̇o₂ (Fig. 8) are not consistent with a requisite role for that factor (pathway) in local metabolic coronary vasodilation. While alternative interpretations certainly remain, we are hopeful that the proposed updated criteria will help facilitate further dialogue and study and assist in better understanding what remains the Gordian knot of the coronary circulation. Given that impaired coronary microvascular function (in the absence of overt atherosclerosis) is now recognized to be a powerful, independent predictor of cardiac morbidity and mortality (83, 86, 101, 102), delineation of the precise mechanisms responsible for the dynamic regulation of coronary blood flow in health and disease is more important than ever.

GRANTS

This work was supported by National Heart, Lung, and Blood Institute Grant HL-136386, Dutch Heart Foundation Grants 2000D038 and 2000D042, and The Netherlands CardioVascular Research Initiative (with support of the Dutch Heart Foundation Grants CVON2014-11 (RECONNECT). Additional support was provided by the use of resources and facilities at the Indiana University School of Medicine (Indianapolis, IN) and the Harry S. Truman Memorial Veteran’s Hospital (Columbia, MO).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

J.D.T., A.K., H.E.B., S.B.B., D.M., and D.J.D. analyzed data; J.D.T., A.G.G., A.K., H.E.B., S.B.B., D.M., and D.J.D. interpreted results of experiments; J.D.T. and A.G.G. prepared figures; J.D.T. and D.J.D. drafted manuscript; J.D.T., A.G.G., A.K., H.E.B., S.B.B., D.M., and D.J.D. edited and revised manuscript; J.D.T., A.G.G., A.K., H.E.B., S.B.B., D.M., and D.J.D. approved final version of manuscript.