Effects of impaired microvascular flow regulation on metabolism-perfusion matching and organ function

Abstract

Impaired tissue oxygen delivery is a major cause of organ damage and failure in critically ill patients, which can occur even when systemic parameters, including cardiac output and arterial hemoglobin saturation, are close to normal. This review addresses oxygen transport mechanisms at the microcirculatory scale, and how hypoxia may occur in spite of adequate convective oxygen supply. The structure of the microcirculation is intrinsically heterogeneous, with wide variations in vessel diameters and flow pathway lengths, and consequently also in blood flow rates and oxygen levels. The dynamic processes of structural adaptation and flow regulation continually adjust microvessel diameters to compensate for heterogeneity, redistributing flow according to metabolic needs to ensure adequate tissue oxygenation. A key role in flow regulation is played by conducted responses, which are generated and propagated by endothelial cells and signal upstream arterioles to dilate in response to local hypoxia. Several pathophysiological conditions can impair local flow regulation, causing hypoxia and tissue damage leading to organ failure. Therapeutic measures targeted to systemic parameters may not address or may even worsen tissue oxygenation at the microvascular level. Restoration of tissue oxygenation in critically ill patients may depend on restoration of endothelial cell function, including conducted responses.

1 |. INTRODUCTION

In caring for the critically ill, clinicians may face the conundrum of a patient with normal ventilation, cardiac output, and arterial oxygen saturation, but with evidence of renal¹ or hepatic² failure. Conventional clinical decision-making in these situations is often based on macroscopic parameters such as heart rate and blood pressure that do not provide insight into conditions at the microvascular level. As a result, therapeutic decisions may not address the underlying pathophysiological derangements leading to organ failure.

Normal tissue function depends on adequate oxygen supply. Although cellular hypoxia can result from defects anywhere along the oxygen transport pathway (pulmonary uptake, blood flow, uptake by mitochondria), deficits not attributable to impairment of overall ventilation or blood flow can result from heterogeneous oxygen transport at the microvascular level. Tissue oxygen levels vary widely over short length scales (tens of microns). Microvascular networks are heterogeneous in structure (diameter, length of flow pathways) and function (flow velocity, oxygen content). This heterogeneity can result in impaired oxygen extraction,³ and regions of hypoxia or anoxia can occur even in tissue that receives an adequate overall oxygen supply.^4,5 This review addresses the role of local regulation of blood flow in overcoming this heterogeneity and matching perfusion to metabolic demand under normal and pathological conditions.

2 |. MICROVASCULAR HETEROGENEITY

2.1 |. Causes of heterogeneity

Oxygen is transported throughout the body by convection in flowing blood and by diffusion from blood into surrounding tissue. The continuous need for oxygen together with its short diffusion distance⁴ necessitates a convective delivery system that places erythrocytes close to every living cell. The network of microvessels that fulfills this requirement is heterogeneous in structure (Figure 1). This heterogeneity, coupled with spatial and temporal variations in flow and demand, requires local regulation of blood flow to prevent regions of hypoxia. Similarly, local regulation of blood flow in the lung is needed to avoid high blood flow to poorly ventilated regions, which would result in impaired blood oxygenation. The importance for tissue oxygenation of ventilation-perfusion matching in the lungs and metabolism-perfusion matching in the systemic circulation is illustrated schematically in Figure 2A and B.

FIGURE 1.

Representative microvascular network derived from observations of rat mesentery illustrating structural heterogeneity. A, Main arteriolar inflow. V: Main venular outflow (reprinted from Roy et al.⁵⁶)

FIGURE 2.

Schematic diagrams indicating how lack of ventilation-perfusion matching or metabolism-perfusion matching can cause hypoxia. A, (a) In the lungs, if poorly ventilated alveoli (indicated by small shaded shape) receive high perfusion, then blood may be poorly oxygenated. Thicknesses of lines represent relative distribution of blood flow. (b) In the systemic circulation, if tissue regions with high metabolic demand by mitochondria (represented by large shaded shape) receive low perfusion, then oxygen supply may be inadequate. B, (c) Redistribution of flow in the lungs, for example, by hypoxic vasoconstriction, reduces flow to poorly ventilated regions, improving overall blood oxygenation. (d) Redistribution of flow in peripheral circulation, for example, by local metabolic regulation of blood flow, increases flow to regions of high metabolic demand, improving tissue oxygenation (represented by large unshaded shape). C, Effects of heterogeneous capillary flow rates on oxygen delivery. Oxygen saturation in arteriole A is set to 1. In the homogeneous case, capillaries C1-C4 all have flow of 2.5 (arbitrary units). Saturation in venule V is 0.2, that is, 80% extraction. In the heterogeneous case, the same flow is distributed (1, 2, 3, 4) to capillaries C1-C4. Mixed saturation in V is 0.3 (dashed line), that is, 70% extraction. C1 is anoxic along its downstream half, implying tissue hypoxia. For simplicity, oxygen delivery per unit length is held constant if saturation is above zero

2.1.1 |. Structural heterogeneity

Microvascular network structures are not genetically predetermined, but arise from angiogenesis,^6,7 a stochastic process involving vessel growth in response to metabolic signals and pruning of redundant vessels. This leads to variations in vessel length and diameter.^8–12 An additional source of heterogeneity is the “dimensional problem”.¹³ A three-dimensional delivery system can provide spatially uniform convective transport to a two-dimensional region by means of symmetrically branching networks, all flow pathways being equivalent in geometry and flow (Figure 3A). If, however, a three-dimensional region must be supplied, then the branching network is embedded within the region itself, which results in non-equivalent flow pathways. This problem is exacerbated when feeding and draining vessels run adjacent to each other, as is often the case (Figure 3B). A high degree of heterogeneity among flow pathways is inevitable.

FIGURE 3.

Schematic illustrating the “dimensional problem.” A, Network supplying a two-dimensional region. B. Network supplying a three-dimensional region

2.1.2 |. Flow heterogeneity

According to Poiseuille’s law, flow resistance of a blood vessel is proportional to segment length and inversely proportional to the fourth power of diameter.¹¹ The structural heterogeneity of microvascular networks thus results in wide variations in flow rates.^14–18 A further source of heterogeneity is the particulate nature of blood.¹⁹ At diverging bifurcations, erythrocytes preferentially enter the branch with higher flow (phase separation), leading to wide variations in microvessel hematocrit.^11,20,21 With hemodilution,²² anemia, or other pathophysiological conditions, some microvessels may receive few or no red blood cells, leading to hypoxia.²³ Conditions resulting in lowered hematocrit such as hemorrhagic shock can lead to increased temporal and spatial heterogeneity, particularly following fluid resuscitation.^24,25 Spontaneous oscillations in vessel diameter, termed vasomotion, can redistribute blood flow and may affect oxygen transport.^26,27 As measured by Poole and colleagues in skeletal muscle using phosphorescence quenching, a substantial oxygen gradient exists between the microvasculature and the interstitium,^28,29 highlighting the importance of a homogeneous red blood cell distribution in maintaining adequate supply to tissues with high demand.

The effect of heterogeneity on hemodynamic parameters can be quantified in terms of capillary transit time heterogeneity (CTH), defined as the standard deviation of the transit time distribution,³⁰ which increases in pathophysiological conditions. Alternatively, capillary outflow saturation heterogeneity has been utilized as an index of functional heterogeneity.^31,32 Figure 2C illustrates how heterogeneity in capillary flow rates results in heterogeneous capillary outflow saturation, tissue hypoxia, and reduced oxygen extraction. Capillary flow in vivo is typically much more heterogeneous than shown in this example.⁸ In this schematic representation, capillaries are shown as having equal lengths for simplicity. In reality, flow pathways are heterogeneous in length, and heterogeneous outflow saturation can be caused by poor matching of perfusion to metabolic demand. This representation does not include the effect of oxygen diffusion from arterioles to tissue, which affects oxygen delivery by capillaries and can, for example, result in reverse oxygen diffusion from tissue into capillaries.^33,34

2.1.3 |. Demand heterogeneity

In many organs, spatial and temporal changes in demand necessitate concomitant changes in blood flow to maintain oxygen availability. One example is differential activation of muscle fiber types during exercise.^35,36 A recent computational model accounting for spatial heterogeneity in types and sizes of muscle fibers demonstrated that fiber size heterogeneity was a primary cause of local hypoxia,³⁷ but that non-equilibrium states and high demand conditions such as heavy exercise could exacerbate local discrepancies between supply and demand. A biphasic pattern of oxygen delivery relative to consumption has been demonstrated in skeletal muscle, with increased heterogeneity at low-to-moderate exercise intensities but a decrease at high exercise intensity corresponding to more uniform capillary perfusion.^38,39 Another example is increased regional cerebral metabolic activity, which can occur due to localized neuronal activation. Increased oxygen demand is matched by increased blood flow via neurovascular coupling, a process that relies on multiple cell types and signaling pathways. Dysfunction in neurovascular coupling resulting in hypoxia is suspected in the pathophysiology of various diseases including vascular dementia.⁴⁰

2.2 |. Role of flow regulation in metabolism-perfusion matching

Active vascular responses are needed to counteract the inherent heterogeneity of microvascular flow and ensure adequate tissue oxygenation. These responses can be classified according to the time scales over which they act. Over time scales of hours to weeks, all blood vessels, from capillaries to arteries, are subject to structural adaptation. This occurs during growth and development and in response to changing functional needs⁶ and mitigates the intrinsic heterogeneity of microvascular perfusion.¹³ Over time scales of seconds to minutes, flow regulation is responsible for modulation of local perfusion.⁴¹ Contraction and relaxation of vascular smooth muscle in small arteries and arterioles cause diameter changes that redirect flow via changes in flow resistance.

Vascular smooth muscle tone is modulated by local concentrations of vasoactive metabolites and mediators, autonomic influences, and hemodynamic factors, and conducted responses from downstream vessels.⁴² Hemodynamic factors affecting vessel diameter include responses to wall shear stress and circumferential wall stress generated by transmural pressure. Increases in wall shear stress are sensed by endothelial cells and typically result in vasodilation (decreased vascular tone) due to the release of mediators including NO, prostaglandins, and EDHF (endothelium-derived hyperpolarizing factor). Increases in transmural pressure activate mechanosensitive ion channels in vascular smooth muscle leading to vasoconstriction (increased vascular tone), termed the myogenic response.^43,44 Sympathetic stimulation causes vasoconstriction and serves to direct flow to skeletal muscle during exercise.^45–47

2.2.1 |. Metabolic signals

The mechanisms by which metabolic needs are signaled to the vasculature are not well understood. Multiple signaling mechanisms have been proposed, and more than one mechanism may be active in any given situation. Sensing of oxygen levels is typically involved. In the brain, however, metabolic flow regulation is considered to be independent of oxygen levels.⁴⁸ Metabolic signals may originate in erythrocytes, in vessel walls, or in surrounding tissue.⁴¹ One mechanism of oxygen sensing involves the release of ATP by erythrocytes at a rate that depends on saturation, and ATP binding to receptors on endothelial cells.^49–51 Erythrocyte-dependent NO vasodilator activity has also been implicated as a potential mechanism.⁵² Endothelial cells release metabolites under hypoxic conditions, including NO, prostaglandins, EDHF, and adenosine. Possible metabolites arising from tissue include carbon dioxide (resulting in decreased pH), and breakdown products of ATP including adenosine.⁵³ Neural activity causes increases in extracellular potassium levels. This may be an important metabolic signal in neurovascular coupling and may also play a role in initiating vasodilation when skeletal muscle is stimulated. Another possible mechanism involves a direct effect of hypoxia on smooth muscle cell function causing vasodilation.⁵⁴ The differing functional ranges of oxygen tension at which these mechanisms operate may imply that their effect varies depending on local conditions. The relative roles of signals derived from erythrocytes, vessel walls, and parenchyma are not known.⁵⁵ Theoretical arguments imply that erythrocyte-derived signals alone cannot provide adequate flow regulation in heterogeneous networks.⁵⁶ Experiments have demonstrated the role of non–erythrocyte-dependent mechanisms.⁵⁷ A similar conclusion applies with regard to the role of erythrocyte-derived signals in structural adaptation.^41,58

2.2.2 |. Role of conducted responses

The blood flow rate in any vessel depends on the resistance not only of that segment but also of the flow pathway feeding and draining it. The largest component of overall flow resistance resides in the arterioles. Modulation of capillary flow therefore requires control of the diameters of upstream arterioles.⁵⁹ This is achieved largely by conducted responses propagated along vessel walls.^60,61 Conducted responses involve changes in cell membrane potential and/or intra-cellular ion concentrations, which are transmitted via gap junctions along the endothelial cell layer over distances on the scale of mm⁶² and between endothelial and vascular smooth muscle cells.⁶³ Gap junction connexins involved in conducted responses include Cx37, Cx40, and Cx43 in endothelium and Cx37, Cx40, Cx43, and Cx45 in vascular smooth muscle.⁶⁴ Although other mechanisms exist for co-ordinating vascular constriction and dilation along flow pathways,⁶⁵ upstream conducted responses play a critical role in both flow regulation^63,66 and structural adaptation.⁶ These findings suggest that impairment of conducted responses may play a role in the poor tissue oxygenation observed in pathophysiological conditions, even when perfusion is adequate.⁶⁷

3 |. IMPAIRED METABOLISM-PERFUSION MATCHING: IMPLICATIONS FOR OXYGEN DELIVERY AND UPTAKE

3.1 |. Mechanisms and consequences of impaired flow regulation

Failure of local flow regulation can lead to local areas of inadequate oxygen delivery and impaired oxygen extraction.³ Acute and chronic pathophysiological conditions such as sepsis, peripheral vascular disease, and myocardial ischemia can impair regulatory mechanisms^68,69 and result in local heterogeneity leading to hypoxia and potential organ failure. Potential modes of dysregulation include endothelial cell dysfunction^70–72 and loss of endothelial cell coupling causing attenuation of conducted signals.⁷³ Examples of clinical conditions resulting in increased microvascular perfusion heterogeneity and organ dysfunction are discussed below, followed by organ-specific considerations.

3.2 |. Clinical implications of impaired flow regulation

3.2.1 |. Sepsis

Sepsis results from a dysregulated systemic inflammatory response to an infectious pathogen.⁷⁴ Global vasodilation resulting from inflammation results in maldistribution of oxygen delivery and impaired oxygen extraction.⁷⁵ The syndrome has been defined based on a set of clinical criteria.⁷⁶ If not treated aggressively within an appropriate time, it can lead to hypoperfusion, organ failure, and death.⁷⁷ Failure of one or more organ systems (most commonly respiratory, cardiovascular, and renal) correlates with mortality, and survivors often require continued care.⁷⁸

Features of sepsis include a decrease in systemic vascular resistance (distributive shock) and impaired oxygen extraction despite increased cardiac output and increased systemic oxygen delivery.⁷⁷ Although cardiac output is often preserved or even increased following fluid resuscitation, sepsis can also cause reversible depression in left ventricular ejection fraction and diastolic dysfunction,⁷⁹ leading to superimposed cardiogenic shock. Various classes of sepsis biomarkers have been proposed and used for prognostication,⁸⁰ reflecting the variable presentation and course of this condition.

At the microcirculatory level, derangements in microvascular flow regulation, erythrocyte deformability, leukocyte and platelet activation, and microvascular permeability result in increased heterogeneity of microvascular flow and oxygen delivery, which in combination with increased oxygen demand and mitochondrial dysfunction can result in regional hypoxia and organ failure.^81–83 Furthermore, inflammation and increased capillary permeability can cause hypovolemic shock and tissue edema as fluid leaks out of the microcirculation⁸⁴; this can be exacerbated by damage to the glycocalyx.⁸⁵ In resistance vessels, altered vascular reactivity⁸⁶ due to increased release of vasoactive substances such as adenosine and nitric oxide^87,88 along with impairment of autonomic tone⁸⁹ and conducted responses result in altered patterns of blood flow including shunting.⁹⁰

The term shunting refers to partial diversion of oxygen delivery away from target tissues, which can result in inadequate oxygen utilization.⁹¹ It is characterized by values of venous PO₂ higher than microcirculatory PO₂, and this difference (termed the PO₂ gap) has been observed in hemorrhagic shock.⁹² Anatomical shunting via collaterals has been observed in several organs.^93–95 Functional shunting can result from (i) diffusive transfer of oxygen from arterioles to venules^33,96,97; (ii) failure of flow regulation and increased flow heterogeneity (the focus of this review); or (iii) the theoretical inability of Hb to unload rapidly enough as it passes through the capillaries due to rapid transit times.^98,99 The poor matching of perfusion and oxygen delivery to metabolic needs is a hallmark of sepsis and a likely cause of tissue damage and organ failure.⁹¹ A classic study investigating blood flow under conditions of sepsis and endotoxemia using microspheres in dogs showed little evidence of systemic shunting but demonstrated the presence of splanchnic and renal shunting.¹⁰⁰ As illustrated in Figure 2, heterogeneity in capillary flow can lead to an oxygen extraction deficit, suggesting the use of vasodilators to open up these “weak microcirculatory units”.¹⁰¹ Alternatively, the use of oxygen carriers including perfluorocarbons¹⁰² and hemoglobin-based oxygen carriers to improve transport from these vessels has been investigated.^92,103

The role of mitochondrial dysfunction in sepsis also remains unclear. The presence of impaired organ function in sepsis in the face of adequate oxygen delivery naturally leads to impaired cellular respiration as a possible explanation. Singer and colleagues assessed muscle biopsies for levels of ATP along with markers of mitochondrial activity, NO production, and oxidative stress in critically ill patients relative to controls¹⁰⁴ and found that shock severity (as measured by pressor requirement) correlated with mitochondrial dysfunction. Impaired oxygen extraction as a result of impaired cellular respiration has been termed cytopathic hypoxia.¹⁰⁵

Several other mechanisms have also been implicated in the impaired oxygen extraction seen in sepsis. Heterogeneity of perfusion^75,90 can lead to impaired metabolism-perfusion matching.¹⁰⁶ Aggregation and plugging of microvessels due to altered erythrocyte and leukocyte rheology^77,90,107 can occur along with vascular microthrombosis.¹⁰⁸ Decreased sensitivity of adrenergic receptors to norepinephrine with acidosis and hypoxia can lead to impairment of the autonomic regulation that contributes to metabolism-perfusion matching. Also, in spite of increased sympathetic activity,¹⁰⁹ widespread vasodilation may occur due to various mechanisms, including adenosine⁸⁸ and increased NO production due to upregulation of iNOS,¹¹⁰ causing decreased whole body and regional extraction. These characteristics are consistent with the paradigm of sepsis as a disease of the microcirculation.^81,111–113 Since the severity of microcirculatory dysfunction is associated with poor outcome in critical illness,^114,115 several interventions including fluid resuscitation,^116–118 vasoactive agents,^115,119,120 and NOS inhibition¹¹⁰ have been proposed and utilized to optimize oxygen delivery at the microcirculatory level.^84,121

Mathematical models of the microcirculation in sepsis have simulated heterogeneity in tissue perfusion by assuming populations of capillaries with stopped flow, and have demonstrated its detrimental effect on oxygen transport.^122,123 As models increase in sophistication, the ability to quantitatively characterize the mechanisms of decreased extraction and response to interventions may allow optimization of pharmacologic management (ie, choice of vasopressors) to reverse maldistribution of flow in vasodilatory states.

3.2.2 |. Metabolic syndrome

A recurring theme in microvascular dysfunction is the role of oxidative stress. Metabolic syndrome represents a constellation of clinical conditions including obesity, hypertension, dyslipidemia and atherosclerosis, and impaired glycemic control that is associated with chronic inflammatory and prothrombotic states.¹²⁴ The underlying pathophysiology of metabolic syndrome is thought to originate with impaired vascular reactivity, decreased perfusion, and microcirculatory dysfunction, causing peripheral vascular disease (PVD) and affecting the brain, heart, and skeletal muscle. Although many pathways are involved in the generation of reactive oxygen species (ROS) in affected individuals, a major mechanism is the uncoupling of nitric oxide production by eNOS so that superoxide is formed instead. This occurs when necessary cofactors are unavailable and/or peroxynitrite (generated by NO scavenging) is present.¹²⁵ The presence of superoxide along with other ROS (such as hydrogen peroxide) and reactive nitrogen species interferes with the regulation of vascular tone. Overproduction of NO can also lead to endothelial dysfunction by causing peroxynitrite formation. The resultant oxidative stress leads to a blunting of the vasodilatory response in hypoxia due to increased endogenous vasoconstrictor production via the cyclooxygenase pathway.¹²⁶ These changes manifest as structural and functional alterations in the vascular wall in affected patients, leading to limitations in skeletal muscle and cerebral blood flow. In an animal model of metabolic syndrome (the obese Zucker rat), cerebral arteries were found to be stiffer and narrower with impaired vasodilatory capacity and decreased nitric oxide availability relative to controls.¹²⁴ Similarly (in a range of rat models with increasing risk), vascular reactivity and NO bioavailability were found to be progressively more impaired with increasing PVD risk. Additionally, the perfusion distribution coefficient (a measure of temporal perfusion heterogeneity at arteriolar bifurcations) was found to be higher in higher-risk rats and trended toward affecting larger vessels as risk increased.¹²⁷ Finally, in a study using the obese Zucker rat, Frisbee and colleagues investigated the nature of vascular dysfunction by using adrenergic stimulation in addition to pressor response and endothelium-dependent vasodilation; blocking the adrenergic response to simply improve blood flow did not improve oxygen uptake and muscle performance as much as also restoring endothelial function,¹²⁸ suggesting distinct mechanisms of dysfunction affecting different calibers of vessels.¹²⁹ Modeling results for skeletal muscle confirm that perfusion heterogeneity adversely affects tissue oxygenation in the metabolic syndrome.¹³⁰

3.2.3 |. Diabetes mellitus

Diabetes is characterized by high glucose levels either due to inadequate insulin production (T1D) or due to insulin resistance in peripheral tissues (T2D) as seen in metabolic syndrome.¹³¹ Many of its common complications (retinopathy, nephropathy, neuropathy,¹³² myocardial dysfunction) are vascular in origin and result from oxidative stress and inflammation. Abnormal angiogenesis results from increased VEGF (vascular endothelial growth factor) levels in the retina and kidneys, and may also destabilize vascular walls.¹³³ Advanced glycation end products (AGEs) and microRNAs have been implicated in the development of vascular disease in T2D.¹³⁴ Walls of large vessels show structural changes secondary to inflammation (cell proliferation, hypertrophy), with decreased compliance. Both small and large vessels show impaired endothelium-dependent vasodilation, increased endothelial permeability, and oxidative stress. In an animal model of T2D, the observed decreased vascular reactivity and blunted active hyperemia were attributed to increased thromboxane production and oxidative stress¹³¹; muscle perfusion was partially restored with antioxidant and anti-thromboxane therapy. The significant role that oxidative stress and inflammation play in diabetic microangiopathy suggests that targeting these mediators and controlling comorbid conditions such as hypertension may be as important as normalizing glucose levels.¹³⁴ This is corroborated by differences in functional abnormalities of the microvasculature in T1D vs T2D.¹³⁵ Finally, diabetes may influence muscle glucose uptake via a direct effect on muscle perfusion,¹³⁶ although this remains controversial.¹³⁷

3.3 |. Organ-specific considerations

3.3.1 |. Kidney

The kidney receives a disproportionate amount of blood flow for its size and has one of the highest oxygen consumption rates in the body.¹³⁸ Its high metabolic demand required to facilitate sodium transport, and its complex vascular architecture makes it particularly vulnerable to hypoxic injury. The renal cortex receives the majority of renal blood flow (RBF) relative to the medulla and renal PO₂ decreases from the cortex to the medulla as the osmotic gradient increases.¹³⁹ Acute kidney injury (AKI) in the critically ill can result from ischemic or nephrotoxic insults, but is also common in sepsis even among patients who have not experienced hypoxemia or hypotension resulting in decreased RBF and/or oxygen delivery to the kidney.¹⁴⁰ The severity of AKI is characterized by the degree of oliguria, an increase in serum creatine, and the need for renal replacement therapy (KDIGO–Kidney Disease: Improving Global Outcomes Criteria),¹⁴¹ but these are often late signs and highlight the need for early diagnosis and treatment. Since histological studies of AKI particularly in sepsis may not show overt evidence of structural injury such as tubular necrosis,^142,143 focus has shifted to possible proinflammatory and prooxidant effects causing microcirculatory dysfunction and alterations in vascular reactivity. This is corroborated by studies showing that simply targeting higher perfusion pressures does not lead to improvements in oxygenation.¹⁴⁴

The inflammatory response that characterizes sepsis is thought to initiate a cascade culminating in the release of cytokines, chemokines, and other mediators that cause endothelial cell activation and microvascular dysfunction.¹⁴⁵ Consequences include impaired flow regulation due to regional deficits in NO-mediated vasodilation causing heterogeneity in perfusion and oxygen levels, enhanced release of ROS, platelet aggregation, and damage to the glycocalyx and endothelial junctions¹⁴⁶ leading to increased vascular permeability and edema. Regional variations in renal perfusion and oxygen levels were recently observed in renal injury using pimonidazole staining,¹⁴⁷ suggesting that shunting of oxygen might explain the impaired oxygen utilization and extraction seen in AKI.⁹¹ This is supported by a study showing that arteriovenous shunting was an important contributor in maintaining renal PO₂ in the face of changing RBF.¹⁴⁸ Since renal oxygen delivery normally exceeds demand, the possibility that preglomerular shunting could serve to spare the renal parenchyma from exposure to high oxygen levels and ROS was proposed.^149,150 Arteriovenous oxygen transport may be facilitated by the anatomical arrangement of the renal vasculature.^151,152 This concept is supported by mathematical modeling^94,153,154 although recent models^153,155 predict a relatively small shunt fraction under normal conditions. With regard to the role of mitochondria in AKI, a recent in vitro study showed that mitochondrial function was impaired in renal tissue exposed to endotoxin¹⁴⁰ and that mitochondrial damage may occur even in the absence of tubular injury.¹⁵⁶

3.3.2 |. Skeletal muscle

In skeletal muscle, autonomic regulation modulates flow distribution to recruited motor units.³⁵ The baseline diameter of skeletal muscle arterioles is determined in part by resting sympathetic nerve activity (SNA).¹⁵⁷ Increased SNA (eg, during exercise) leads to increased cardiac contractility and arteriolar vasoconstriction, which maintains systemic blood pressure as the skeletal muscle vasculature dilates. The local metabolic vasodilation (termed functional sympatholysis) serves to direct blood to metabolically active areas.^158–160 This interaction is mediated by the differential response of larger vs. smaller arterioles to vasodilatory metabolites^46,161 due to the distribution of α adrenergic receptor subtypes. The sympathetic vasoconstriction of smaller vessels with higher concentrations of α₂ receptors appears to be more susceptible to inhibition by vasodilatory metabolites as compared to larger vessels with higher concentrations of α₁ receptors,^162,163 providing a mechanism for increasing perfusion to metabolically active areas. Attempts to increase oxygen delivery by pharmacologic vasodilation have resulted in either no improvement in maximal aerobic capacity¹⁶⁴ or decreased oxygen extraction.¹⁶⁵ Despite an increase in blood flow, vasodilation via adenosine results in impaired metabolism-perfusion matching (ie, decreased extraction), likely due to shunting of blood to non-exercising tissue.^164,166 Similarly, experiments performed at altitude with adenosine vasodilation show a paradoxical worsening of oxygen uptake with reduced extraction due to shunting.¹⁶⁵ These results indicate that adrenergic tone is important for directing flow to metabolically active areas.^167,168 Implications for impaired vascular reactivity (as seen in metabolic syndrome) include the development of peripheral vascular disease and mitigation of this risk by exercise training.¹⁶⁹

3.3.3 |. Brain

Under normal conditions, increases in regional cerebral metabolic demand either increase extraction¹⁷⁰ or increase perfusion via neurovascular coupling.^171,172 This involves conducted responses that vasodilate upstream arterioles and small arteries.¹⁷¹ Several disorders have been linked to cerebrovascular dysfunction and impaired neurovascular regulation including vascular dementia,⁴⁰ stroke,¹⁷¹ and Alzheimer’s disease.^{171,173–176} Cerebral autoregulation maintains blood flow under a wide range of perfusion pressures.⁴⁰ The cerebral blood supply arises from the Circle of Willis, and the vascular arrangement in the deep white matter is such that the watershed regions between the anterior and middle cerebral artery distributions have poor collateral flow. This area is vulnerable to disruptions in perfusion and can be affected by decreases in global cerebral perfusion and neuropathological small vessel lesions affecting arterioles.¹⁷⁷ Vascular dementia is often functional but is sometimes associated with pathological features including leukoaraiosis, lacunar infarcts, microbleeds, microinfarcts, and cerebral amyloid angiopathy.⁴⁰ In many cases, these lesions are accompanied by attributes of neurodegenerative diseases such as the neurofibrillary tangles and amyloid plaques seen in Alzheimer’s disease. These processes may be synergistic in that vascular insufficiency promotes amyloid deposition and impaired clearance, and amyloid reduces cerebral blood flow and impairs functional hyperemia.⁴⁰ Both vascular dementia and Alzheimer’s disease exhibit vascular dysfunction including impaired reactivity, altered cerebrovascular autoregulation, and disruption of the blood-brain barrier.^174,178 The demyelinated neurons and damaged glia are unable to support the endothelium, leading to capillary rarefaction. The underlying cause of vascular dysfunction is thought to be oxidative stress and inflammation leading to endothelial damage and disruption of neurovascular coupling.^40,176 Since no effective treatment has been identified, efforts have focused on preventive measures such as control of hypertension and other risk factors.⁴⁰ Ischemic strokes result in similar impairment of reactivity and autoregulation¹⁷⁷ and can even result in decreased blood flow and metabolism in intact areas.¹⁷¹ Eventual consequences of impaired cerebral blood flow (as seen in metabolic syndrome for example) can include cognitive decline and stroke.¹²⁴

3.3.4 |. Heart

The myocardium is characterized by regional flow heterogeneity^179,180 and regional variations in oxygen demand.¹⁰ These variations lead to heterogeneity in oxygen levels at the microvascular level.¹⁰ Recent evidence suggests that the fractal nature of myocardial blood flow correlates more with metabolic activity than vascular structure,¹⁸¹ suggesting the strong influence of metabolic flow regulation in the heart. Failure of flow regulation is seen in a condition called microvascular angina (previously referred to as cardiac syndrome X)^182,183 in which patients (women more often than men) experience angina in response to microvascular dysregulation without arterial occlusion¹⁸⁴; these symptoms may or may not be accompanied by ECG changes and/or regional wall motion abnormalities. The pathophysiology involves endothelial and autonomic dysfunction resulting in vasoconstriction and/or inadequate vasodilation in response to metabolic demand.^183,185 The condition is difficult to diagnose since patients typically have normal coronary angiograms. Noninvasive methods such as vasodilator stress cardiac MRI¹⁸⁵ and invasive methods such as coronary flow reserve (CFR) and the index of microcirculatory resistance (IMR) have been used.^183,186,187 Patients with symptoms of microvascular angina are typically treated with drugs used for ischemia such as beta-adrenergic blocking agents, calcium channel blockers, and nitrates¹⁸⁵ as well as preventative measures, since microvascular angina can coexist with atherosclerotic heart disease. Microvascular dysfunction is often associated with atherosclerotic heart disease^183,188 and can persist even after revascularization of large coronary vessels¹⁸⁹; capillaries experience oxidative stress and in some cases exhibit loss of pericytes and rarefaction, limiting blood flow. The systemic implications of the atherogenic, inflammatory, and prothrombotic state seen in the metabolic syndrome include impaired vascular reactivity and peripheral vascular disease, culminating in angina or heart failure from hypertension and increased afterload.¹²⁷

3.3.5 |. Lung

Efficient uptake of oxygen through the alveolar-capillary membrane requires a normal diffusing capacity¹⁹⁰ and intact flow regulation to match perfusion to ventilation.¹⁹¹ ARDS (acute respiratory distress syndrome) is characterized by severe hypoxemia resulting in high mortality among critically ill patients. Its pathogenesis includes pulmonary endothelial injury resulting in increased vascular permeability and pulmonary edema as well as alveolar epithelial injury and increased interstitial fluid.^192–194 Inflammation and accumulation of protein-rich edema fluid impair gas exchange and cause alveolar flooding.¹⁹⁵ The increased diffusion distance across the alveolar-capillary membrane coupled with impaired flow regulation and local or regional heterogeneity of ventilation and perfusion contributes to decreased arterial oxygenation.¹⁹⁶

Lung injury in patients with ARDS is associated with changes in ventilation at the regional and local levels, leading to heterogeneity at multiple scales. Increased heterogeneity is associated with increased dead space fraction.^197,198 Ventilation changes in distal aspects of the tracheobronchial tree can occur due to mucous plugging, particularly in cases of infection. Pulmonary edema due to alveolar epithelial injury can alter alveolar mechanics (preventing alveoli from opening during tidal ventilation); alveolar flooding (an extreme case of increased alveolar-capillary membrane thickness) can prevent ventilation altogether.

In the absence of well-distributed ventilation, adequate matching of perfusion to ventilation through flow regulation is crucial. Hypoxic vasoconstriction (HPV) is a key mechanism for matching perfusion to ventilation.^199–202 Capillary endothelial cells act as the sensors of local oxygen tension²⁰³ along with pulmonary artery smooth muscle cells (PASMC).²⁰⁴ Upstream conduction of this signal via gap junctions in endothelial cells has been proposed as the mechanism for the vasoconstriction of upstream arterioles via contraction of PASMC.²⁰⁵ Endothelial injury and/or decreased Cx40 expression may explain the impairment of HPV and failure of flow regulation in ARDS²⁰⁶ and illnesses such as COVID-19.²⁰⁷ In addition, the cystic fibrosis transmembrane conductance regulator (CFTR) has been shown to be necessary for an intact HPV response, and recent evidence demonstrates that some lung infections block HPV by inhibiting CFTR.²⁰⁸ The extent to which HPV is impaired in ARDS and is attenuated by various agents used in the treatment of critically ill patients remains unclear.¹⁹⁶

Inhaled NO (iNO) and inhaled prostacyclin (PGI2) have been used to improve ventilation/perfusion matching in mechanically ventilated patients^209,210; although these agents can improve oxygen saturation, they have not been shown to decrease mortality.^210,211 Regional changes in blood flow with iNO have been observed in hypoxia and normoxia on imaging.²¹² Other agents that have been used to redistribute blood flow include almitrine, a selective pulmonary vasoconstrictor of nonventilated lung areas,²¹³ and phenylephrine, an α-receptor agonist causing vasoconstriction.^214,215 Further investigation is needed to determine optimal pharmacologic combinations to improve arterial oxygen saturation.

When supplemental oxygen is not sufficient to raise arterial oxygen saturation, patients are placed on positive pressure ventilation (PPV). Mechanical ventilation itself can cause lung injury²¹⁶ and was shown to worsen ventilation-perfusion matching and oxygenation in an animal model of lung injury by decreasing blood flow to poorly ventilated areas during inspiration.²¹⁷ Flow-controlled ventilation has been proposed to attenuate lung injury and improves homogeneity of aeration in dependent portions of the lung.²¹⁸

In summary, local regulation of pulmonary blood flow is crucial for maintaining oxygen saturation in critically ill patients. In ARDS, the combination of impaired flow regulation and impaired diffusing capacity results in severe hypoxemia and requires judicious application of pharmacologic interventions and appropriate ventilator strategies to redistribute blood flow.

3.3.6 |. Tumors

As a tumor develops, angiogenesis is stimulated by its need for oxygen and nutrients, and enables tumors to grow to sizes larger than the diffusion distances for those substances.²¹⁹ Angiogenic activators such as VEGF are instrumental in this process; the expression of VEGF and its receptor is induced by the effects of hypoxia on HIF-1α (hypoxia-inducible factor 1, α subunit).²²⁰ Tumor microvessel networks have an aberrant structure and exhibit increased vascular permeability, resulting in heterogeneous blood flow and oxygenation.^4,221,222 The abnormal conditions in tumors cause hypoxia and affect the transport of anticancer drugs, often limiting their effectiveness.²²³ Strategies to mitigate these effects include limiting stromal development to minimize microvascular compression and improve flow, normalizing the vascular network by using VEGF inhibitors, and using supplemental oxygen or pharmacologic therapies to reduce oxygen consumption to minimize hypoxia.²²⁴

3.4 |. Implications for therapeutic interventions

Optimizing tissue oxygenation and perfusion at the microvascular level in addition to restoring systemic physiological parameters (‘hemodynamic coherence’²²⁵) is the objective of ‘microvascular resuscitation’.^{115,226–228} Several strategies have been proposed and employed to address the spatial and temporal heterogeneities in flow and oxygenation (‘microcirculatory shock’²²⁹) seen at the microvascular level.^230–233 Methods for monitoring the microvascular response to these interventions^234–236 are also discussed.

3.4.1 |. Vasoactive agents

Pharmacologic agents commonly used in critically ill patients include vasopressors (to increase systemic perfusion pressure),^119,120,237 vasodilators (to increase microvascular flow and decrease shunting),^119,238 and inotropes (to increase cardiac output).^119,120 Clinically, these therapies are typically used following intravascular volume resuscitation, but can be used concurrently with fluid^239–241 or in combination with each other.^242,243 Despite multiple clinical trials involving pharmacologic agents, no definitive correlation has been found between achieving systemic hemodynamic endpoints and improved microcirculatory metabolism-perfusion matching.^119,244

Vasoactive drugs differentially affect vessels of different caliber due to differing distributions of adrenergic receptors.^{120,245–247} Many vasopressors (and vasodilators) also have systemic effects.^101,119,248 For example, norepinephrine has significant α-adrenergic properties and causes peripheral vasoconstriction, and also increased cardiac output due to its β-adrenergic effects.^119,120 Phenylephrine (a selective α₁-adrenergic agonist) causes peripheral vasoconstriction and an increase in mean arterial pressure (MAP), but may decrease cardiac output via increased afterload.¹²⁰ Epinephrine increases MAP via α- and β-adrenergic effects on the peripheral circulation as well as cardiac output; β-effects predominate at lower doses.^119,120 Vasopressin causes contraction of vascular smooth muscle and increases sensitivity to catecholamines.¹²⁰

Although some of these drugs preserve or improve systemic perfusion pressure,^249–251 their effect on microvascular perfusion remains unclear,^{119,248,252–254} especially in conjunction with fluid resuscitation. In fact, the use of vasodilators has also been proposed to increase overall microvascular perfusion and reduce stopped-flow capillaries and/or plasma channels.^255,256 Another proposed unconventional therapy is inhaled nitric oxide to stimulate nitrite and S-nitrosothiol production, causing release of NO by erythrocytes and potentially improving flow regulation in sepsis.²⁵⁷ Potential exists for personalized therapy based on patient phenotype (adrenergic receptor subtypes, responses to pharmacologic agents). Similar techniques could be used in other conditions that involve inflammatory responses and heterogeneity of flow. Hemorrhagic shock and free flap perfusion are conditions in which elucidating the microcirculatory response to fluid resuscitation and vasoactive agents would be of interest.

3.4.2 |. Fluid therapy

Intravenous crystalloid administration has long been the mainstay of therapy in patients with sepsis or distributive shock, to restore intravascular volume, maintain stroke volume and cardiac output, and preserve blood pressure in the context of vasodilation and capillary leak. However, the optimal amount and type of fluid remain a subject of debate.^258,259 Judicious fluid resuscitation is the initial component of protocol-based management (“early goal-directed therapy”)²⁶⁰ as formalized in the commonly utilized “Surviving Sepsis” guidelines,^74,261 although these are controversial.²⁶² Various systemic hemodynamic endpoints for fluid resuscitation have been proposed,²⁶³ including central venous pressure, mean arterial pressure, cardiac index, oxygen delivery, oxygen consumption, and mixed venous oxygen saturation. Clinical trials utilizing such endpoints have not reliably demonstrated improved outcomes, however^264,265 implying that restoring systemic hemodynamics may be insufficient to restore oxygen delivery to tissue.²⁶⁶ One possible cause may be endothelial damage due to the effect of crystalloids on the integrity of the glycocalyx, a luminal barrier layer that plays a role in regulation of vascular permeability, immune function, cellular adhesion, and flow regulation.^267–269 Additionally, alterations in plasma viscosity and erythrocyte deformability induced by administered crystalloids and colloids can alter the flow and distribution of blood and cause heterogeneity in oxygen transport.^270–272 In summary, potential detrimental effects from fluid resuscitation include hemodilution (decreased hematocrit), damage to the glycocalyx resulting in increased permeability leading to tissue and pulmonary edema (and therefore increased diffusion distances),¹¹⁶ altered hemorheology, shunting/malperfusion (resulting in plasma channels and poor tissue oxygenation), and venous congestion.^{85,116,273,274}

Risks of under-resuscitation include hypotension, hypoperfusion, and inadequate oxygen delivery leading to organ failure, but over-resuscitation, even if indicated by systemic hemodynamic parameters, may be detrimental.^116,275 Studies of microvascular oxygenation using phosphorescence techniques in rat intestinal microcirculation have demonstrated the preservation of tissue oxygen levels with crystalloid resuscitation.²⁷⁶ Increased flow may improve perfusion and oxygen delivery at the microcirculatory level only up to a point, at which the effect of hemodilution would supervene and result in hypoxia due to redistribution of erythrocytes, the development of plasma channels,²⁷⁷ and decreased oxygen delivery.^22,278 This is consistent with evidence of shunting seen in experimental models.^279,280 Due to the lack of correlation between systemic hemodynamics and microcirculatory flow, the optimal fluid resuscitation endpoint remains unclear²⁷⁴ and individualized therapy may be indicated.^266,281 Observations of microvascular flow have therefore been used to identify candidates for fluid resuscitation²⁸² based on measured microvascular flow parameters.¹¹⁶

3.4.3 |. Blood transfusion

Blood transfusion improves sublingual microcirculatory density (as measured by video microscopy) and oxygenation (as measured by spectrophotometry).^283,284 Administering whole blood or packed red blood cells prevents hemodilution and may increase vascular tone via the myogenic mechanism, while increasing oxygen delivery. Increased delivery has been shown to improve oxygen utilization in patients with poor baseline extraction²⁸⁵ but may be counteracted to some extent by the increase in flow resistance resulting from hemoconcentration.^286,287 Furthermore, less tissue edema would likely result since more administered volume would be retained within the vasculature. Transfusion of blood with altered oxygen affinity may also serve to improve oxygen unloading in peripheral tissues.^288–290

3.4.4 |. Anesthetic agents

Agents used for sedation and anesthesia comprise another class of commonly used drugs in critically ill patients and are known to affect the vasculature.²⁹¹ However, the effects of anesthetic agents on the microcirculation are variable²⁹² and poorly understood. Effects of inhalational and intravenous agents on erythrocyte deformability,^293,294 integrity of the glycocalyx,²⁹⁵ vascular permeability,²⁹⁶ and vascular reactivity to vasoactive agents^297,298 have all been variously reported.²⁹⁹ As an example, results of a recent study evaluating the effect of general anesthesia on attenuation of renal perfusion and oxygen delivery in an ovine model suggest that the use of IV anesthetics may be preferable to volatile agents if subjects are at risk of renal injury.³⁰⁰ Much work remains to be done to define the most appropriate sedative or anesthetic agents for a particular patient.

3.4.5 |. Monitoring

Conventional methods of monitoring the progress of septic patients include so-called “upstream” variables reflecting oxygen delivery (cardiac output, mean arterial pressure) and “downstream” variables reflecting oxygen utilization (mixed venous oxygen saturation,³⁰¹ lactate levels, urine output). Because these measures may not reflect microcirculatory derangement, methods of clinically evaluating the microcirculation have been developed^302–306 (eg, sublingual capnometry, sidestream dark-field video microscopy, orthogonal polar spectral imaging^307–313). While imaging methods are not in widespread clinical use, algorithms to analyze such images^314,315 can provide serial measurements of local flow velocities, the percentage of perfused vessels, and effective vessel density.^316,317 Derived quantities used to monitor therapy include microvascular flow index (MFI),^282,318 perfused vessel density (PVD),⁵⁹ proportion of perfused vessels (PPV), and a heterogeneity index as well as functional measurements of local tissue oxygen tension as obtained by near-infrared spectroscopy (NIRS).³¹⁸ These measurements are often site-specific¹¹⁷ but have been found to reflect conditions at other sites.³¹⁹ The significance and relevant therapeutic endpoints for these measures are still under investigation.^{144,320–323}

4 |. DISCUSSION

When considered at the whole-organ level, the occurrence of tissue hypoxia in a well-perfused tissue may seem paradoxical. However, this paradox is resolved when microvascular-scale behaviors are considered. The microcirculation is inherently heterogeneous in structure, and active vascular responses on short and long time scales are essential to overcome this heterogeneity and provide adequate oxygenation throughout tissues. Any disturbance of flow regulation can cause local hypoxia. In patients with sepsis or other critical illness, flow regulation may be disturbed due to effects of pathophysiological conditions on endothelial cell function. Conducted responses, which play a vital role in flow regulation, require not only the normal function of individual endothelial cells but also maintenance of their connectivity via intact gap junctions. This aspect of endothelial function is likely to be particularly vulnerable to disruption in abnormal states such as inflammation. Therapeutic measures targeted to systemic parameters may not address or may even worsen such dysfunction. The goal of therapy should therefore be restoration of endothelial cell function, including conducted responses. Further investigation of these aspects of microvascular function with theoretical, experimental, and clinical studies could lead to improved therapies for improving oxygen transport and preventing organ failure in critically ill patients.

5 |. PERSPECTIVES

Critically ill patients depend on adequate oxygen availability at the microvascular level to prevent tissue and organ damage, and regions of hypoxia or anoxia can occur even in a tissue that receives an adequate overall convective oxygen supply. Intact local regulation of blood flow is required to counteract the structural and functional heterogeneity of microcirculatory networks and prevent hypoxia. Given that endothelial injury in critical illness can impair the upstream conducted responses that mediate arteriolar vasodilation and that correcting systemic parameters alone may not improve microvascular oxygen delivery, restoring endothelial function and flow regulation may be the key to preventing organ failure in these “diseases of the microcirculation.”