Autonomic Tone

The lungs are richly enervated with integrated somatic and autonomic fibers that originate, traverse through, and end in the thorax. These neuronal networks mediate multiple homeostatic processes through the autonomic nervous system that alter both instantaneous cardiovascular function (e.g., respiratory sinus arrhythmia) and steady state cardiovascular status (e.g., ADH-induced fluid retention). Numerous cardiovascular reflexes are centered within this network. Inflation induces immediate changes in autonomic output. The most commonly described inflation-chronotropic responses act through vagal-mediated reflex arcs.^41,42 Lung inflation to normal tidal volumes (<10 mL/kg) increases heart rate via parasympathetic tone withdrawal. Inspiration-associated cardioacceleration is referred to as respiratory sinus arrhythmia (RSA)⁴³ and denotes normal autonomic tone.⁴⁴ Loss of RSA is associated with dysautonomia, and its reappearance precedes the return of peripheral autonomic control in diabetics with peripheral neuropathy.⁴⁵ However, some degree of respiratory-associated heart rate change is intrinsic to the heart itself. For example, in denervated human hearts (transplants), a small degree of ventilation-associated changes in heart rate persists,⁴⁶ suggesting that mechanoreceptors in the right atrium can alter sinoatrial tone. Bernardi et al.⁴⁶ demonstrated that this heart rate variability in cardiac transplant recipients had a periodicity twice that of the ventilatory cycle, suggesting that both increases and decreases in venous return and ventricular loading impart some RSA. In support of this concept, Pinna et al. documented that most of the heart rate and arterial pressure changes seen during breathing in patients with severe congestive heart failure (CHF) are more reflective of changes in intrathoracic blood volume than of alterations in autonomic input.⁴⁷

Lung inflation to larger tidal volumes (>15 mL/kg) decreases heart rate. Pulmonary vasoconstriction may occur through vagal reflex arcs⁴⁸ but does not appear to induce significant hemodynamic effects. Reflex arterial vasodilatation can also occur with lung hyperinflation.^41,49–53 This inflation-vasodilatation response appears to be mediated by afferent vagal fibers, because it is abolished by selective vagotomy. Interestingly, blocking sympathetic afferent fibers also blocks this reflex,^51,54 presumably by withdrawing central sympathetic tone. Although this inflation-vasodilatation response induces expiration-associated reductions in LV contractility in healthy volunteers⁵⁵ and in ventilator-dependent patients with the initiation of high-frequency ventilation⁴¹ or hyperinflation,⁵¹ its clinical significance in other patient groups is unknown. Since patients with ALI often ventilate only a relatively small amount of their lungs, the potential exists that these patients may experience regional hyperinflation and may develop reflex cardiovascular depression. Interestingly, several studies comparing larger tidal volume ventilation with pressure-limited ventilation document better hemodynamic status with pressure-limited ventilation. Similarly, although humoral factors (including compounds whose production is dependent on cyclooxygenase activation⁵⁶) released from pulmonary endothelial cells during lung inflation may also induce this depressor response,^57–59 these interactions do not appear to grossly alter cardiovascular status.⁶⁰ In fact, unilateral lung hyperinflation (unilateral PEEP) does not appear to influence systemic hemodynamics.⁶¹ Interestingly, increased levels of nitric oxide (NO) in the exhaled gas of rabbits ventilated at increasing tidal volumes have been reported.⁴⁸ Importantly, for the same decrease in cardiac output, heart rate increases less with the application of PEEP than with hemorrhage.⁴⁸ The reasons for this difference are unknown but may reflect PEEP-induced sympatholytic actions and increased arterial pressure minimizing baroreceptor stimulation.

Both positive-pressure ventilation and sustained hyperinflation stimulate endocrinological responses that induce fluid retention via right atrial stretch receptors. Plasma norepinephrine, plasma renin activity,^62,63 and atrial natriuretic peptide (ANP)⁶⁴ increase during positive-pressure ventilation with or without PEEP. Interestingly, when subjects with CHF are given nasal CPAP, plasma ANP activity decreases in parallel with improvements in blood flow.^65,66

Determinants of Pulmonary Vascular Resistance

Changes in lung volume are caused by changes in transpulmonary distending pressure, the pressure difference between alveolar pressure and ITP. Since pulmonary tissue pressure and ITP are nearly identical, increasing lung volume increases the difference between alveolar and tissue pressures, making pulmonary vascular resistance increase independent of any effect of volume change on humoral or autonomic responses.^18,67–72

Lung inflation primarily affects cardiac function and cardiac output by altering right ventricle (RV) preload and afterload.⁷² RV afterload is the maximal RV systolic wall stress during contraction⁷³ which, by Laplace’s law, equals the product of the RV radius of curvature (a function of end-diastolic volume) and transmural pressure (a function of systolic RV pressure).⁷⁴ Changes in ITP that occur without changes in lung volume, as may occur with obstructive inspiratory efforts, will not alter the pressure gradient between the RV and pulmonary artery nor result in change of pulmonary vascular resistance. Thus, neither straining at stools (Valsalva maneuver) nor obstructive inspiratory efforts (Mueller maneuver) primarily affect RV afterload. Although obstructive inspiratory efforts are usually associated with increased RV afterload, the reason for this effect is backward LV failure and reactive hypoxic pulmonary vasoconstriction.^75,76

Systolic RV pressure approximates transmural systolic pulmonary artery pressure (Ppa) when no pulmonary stenosis is present. Transmural Ppa can increase by one of two mechanisms: (1) an increase in pulmonary arterial pressure without change in pulmonary vasomotor tone, as occurs with increases in blood flow (exercise) or passive increases in outflow pressure (LV failure), or (2) an increase in pulmonary vascular resistance. Usually any increase in transmural Ppa during positive-pressure ventilation is due to an increase in pulmonary vascular resistance, because neither instantaneous cardiac output⁷⁷ nor LV filling¹⁴ changes. Increases in transmural Ppa impedes RV ejection,⁷⁸ decreasing RV stroke volume⁷⁹ and causing RV dilation and passive obstruction to venous return,^56,58 which may rapidly progress to acute cor pulmonale.⁸⁰ If RV dilation and pressure overload persist, RV free wall ischemia and infarction can develop.⁸¹ Importantly, rapid fluid challenges in the setting of acute cor pulmonale can precipitate profound cardiovascular collapse due to excessive RV dilation, RV ischemia, and compromised LV filling through the process of ventricular interdependence. During normal end-inspiration, mild hypoxemia (PaO₂ > 65 mm Hg) and low levels of PEEP (<7.5 cm H₂O) should minimally increase transmural Ppa. If slight increases in transmural Ppa are sustained, however, fluid retention occurs, either by intrinsic humoral mechanisms (increased ANP secretion) or by therapeutic intravascular volume infusion,⁸² resulting in an increase in RV end-diastolic volume maintaining cardiac output.^74,83

The mechanism by which ventilation alters pulmonary vasomotor tone is complex. If regional alveolar PO₂ (PAO₂) decreases below 60 mm Hg, local pulmonary vasomotor tone increases, reducing local blood flow.⁸⁴ This process of hypoxic pulmonary vasoconstriction is mediated in part by variations in the synthesis and release of NO by pulmonary vascular endothelial cells. The pulmonary endothelium normally synthesizes a basal low amount of NO, a potent vasodilator, using endothelial nitric oxide synthase. This basal NO release is highly regulated and dependent on O₂ and is inhibited by both hypoxia and acidosis. If O₂ becomes scarce, NO is not made, and pulmonary vasomotor tone increases to the level that would exist if no NO were present.

Many pulmonary pathologic processes are associated with regional reductions in PAO₂, such as atelectasis, airway obstruction, and ventilation/perfusion mismatching. Hypoxic pulmonary vasoconstriction, by reducing pulmonary blood flow to those hypoxic regions, optimizes ventilation/perfusion matching. However, if alveolar hypoxia occurs throughout the lungs, overall pulmonary vasomotor tone increases, raising pulmonary vascular resistance and impeding RV ejection.⁷³ At low lung volumes, alveoli spontaneously collapse as a result of loss of interstitial traction and closure of the terminal airways, causing alveolar hypoxia. Patients with acute hypoxemic respiratory failure have small lung volumes.^85,86 Therefore, pulmonary vascular resistance is often increased in these patients owing to alveolar collapse and resultant hypoxic pulmonary vasoconstriction.

Mechanical Ventilation-Induced Changes in Pulmonary Vascular Resistance

Mechanical ventilation may reduce active pulmonary vasomotor tone by one of several related processes. Hypoxic pulmonary vasoconstrictor tone may be decreased by increasing global PAO₂ by enriching alveolar gas O₂,^87–90 reexpansion of collapsed alveolar units thereby increasing PAO₂ in those local alveoli,^5,91–93 increased alveolar ventilation and reversal of acute respiratory acidosis,⁹⁰ or merely through decreasing central sympathetic output by allowing the patient in ALI to not fight for every breath.^94,95 Similarly, these effects need not require positive-pressure breaths as much as expansion of collapsed alveoli.⁹⁶ Such recruitment of lung units is usually accomplished by the addition of PEEP or CPAP. Thus, if PEEP opens collapsed lung units and replenishes alveolar gas with O₂, hypoxic pulmonary vasoconstriction will be reduced, pulmonary vascular resistance will decrease, and RV ejection will improve.

Changes in lung volume can also profoundly alter pulmonary vasomotor tone by passively compressing the alveolar vessels.^85,92,93 Pulmonary circulation can be separated into two groups of blood vessels depending on what pressure surrounds them⁹² (Figure 47-4). The small pulmonary arterioles, venules, and alveolar capillaries sense alveolar pressure as their surrounding pressure and are referred to as alveolar vessels. The large pulmonary arteries and veins, as well as the heart and intrathoracic great vessels of the systemic circulation, sense interstitial pressure or ITP as their surrounding pressure and can be called extraalveolar vessels. Alveolar pressure minus ITP is the transpulmonary pressure. Increasing lung volume requires a rise in transpulmonary pressure, so the extravascular pressure gradient between alveolar and extraalveolar vessels varies proportionally with changes in lung volume. Importantly, the radial interstitial forces of the lung that keep the airways patent^91,97,98 also act upon the extraalveolar vessels. As lung volume increases, the radial interstitial forces increase, increasing the diameter of both extraalveolar vessels and airways and resulting in a reduction in airway resistance at higher lung volumes, as well as increased extraalveolar vessel diameter and capacitance.⁹⁹ This tethering effect is lost with lung deflation, thereby increasing pulmonary vascular resistance.^88,91 The collapse of small airways also induces alveolar hypoxia. Thus, at small lung volumes, pulmonary vascular resistance is increased owing to the combined effect of hypoxic pulmonary vasoconstriction and extraalveolar vessel collapse.

Figure 47-4 Schematic diagram of the relation between changes in lung volume and pulmonary vascular resistance, where the extraalveolar and alveolar vascular components are separated. Note that pulmonary vascular resistance is minimal at resting lung volume or functional residual capacity (FRC). As lung volume increases toward total lung capacity (TLC) or decreases toward residual volume (RV), pulmonary vascular resistance also increases. However, the increase in resistance with hyperinflation is due to increased alveolar vascular resistance, whereas the increase in resistance with lung collapse is due to increased extraalveolar vessel tone.

Increases in lung volume progressively raise alveolar vessel resistance, becoming noticeable above resting lung volume or functional residual capacity (FRC).^88,100 There are two causes of the increased alveolar vessel resistance. First, the heart and extraalveolar vessels sense ITP as their surrounding pressure, whereas the alveolar vessels sense alveolar pressure as their surrounding pressure, so an extralumenal transpulmonary pressure gradient exists between extraalveolar and alveolar vessels. As lung volume increases, the extralumenal pressure difference increases as well. If transpulmonary pressure increases enough to exceed intralumenal vascular pressure, the pulmonary vasculature will collapse where extraalveolar vessels pass into alveolar loci, reducing the vasculature cross-sectional area and increasing pulmonary vascular resistance. Similarly, increasing lung volume by stretching and distending the alveolar septa may also compress alveolar capillaries, although this mechanism is less well substantiated. Hyperinflation can create significant pulmonary hypertension and may precipitate acute RV failure (acute cor pulmonale)¹⁰¹ and RV ischemia.⁸¹ Thus PEEP may increase pulmonary vascular resistance if it induces overdistention of the lung above its normal FRC. Recently the effect of inflation on RV input impedance was validated in humans, using echocardiographic techniques.¹⁰² Similarly, if lung volumes are reduced, then increasing lung volume back to baseline levels by the use of PEEP will decrease pulmonary vascular resistance by reversing hypoxic pulmonary vasoconstriction.¹⁰³

Ventricular Interdependence

Changes in RV output must invariably alter LV filling because the two ventricles are serially linked through the pulmonary vasculature. However, LV preload can also be directly altered by changes in RV end-diastolic volume.¹⁰⁴ If RV volume increases, LV diastolic compliance will decrease by the mechanism of ventricular interdependence.¹⁰⁵ Ventricular interdependence functions through two separate processes. First, increasing RV end-diastolic volume will induce a shift of the intraventricular septum into the LV, thereby decreasing LV diastolic compliance¹⁰⁶ (Figure 47-5). For the same LV filling pressure, RV dilation will decrease LV end-diastolic volume and therefore cardiac output. This interaction is believed to be the major determinant of the phasic changes in arterial pulse pressure and stroke volume seen in tamponade, referred to as pulsus paradoxus. Spontaneous inspiration increases venous return, causing RV dilation and decreasing LV end-diastolic compliance. Maintaining a relatively constant rate of venous return, either by volume resuscitation¹⁰⁷ or vasopressor infusion,⁴ will minimize this effect. Thus, the presence of pulse paradoxus can be used as a marker of functional hypovolemia, even if actual intravascular volume status is not reduced.

Figure 47-5 Schematic diagram of the effect of increasing right ventricular (RV) volumes on the left ventricular (LV) diastolic pressure-volume (filling) relationship. Note that increasing RV volumes decrease LV diastolic compliance such that a higher filling pressure is required to generate a constant end-diastolic volume.

(After Taylor RR, Covell JW, Sonnenblick EH, Ross J Jr. Dependence of ventricular distensibility on filling of the opposite ventricle. Am J Physiol 1967;213[3]:711-718.)

Mechanical Heart-Lung Interactions

With hyperinflation, the heart may be compressed between the two expanding lungs,¹⁰⁸ increasing juxtacardiac ITP. The chest wall and diaphragm can move away from the expanding lungs, whereas the heart is trapped within its cardiac fossa, so juxtacardiac ITP may increase more than lateral chest wall or diaphragmatic ITP.^6,16 This compressive effect of the inflated lung can be seen with either spontaneous¹⁰⁹ or positive pressure–induced hyperinflation.^97,98 This decrease in “apparent” LV diastolic compliance¹⁰⁶ was previously misinterpreted as impaired LV contractility, because LV stroke work for a given LV end-diastolic pressure or pulmonary artery occlusion pressure is decreased.^110,111 However, numerous studies have shown that when patients are fluid resuscitated to return LV end-diastolic volume to its original level, both LV stroke work and cardiac output also returned to their original levels^69,107 despite the continued application of PEEP.¹¹²

Takata et al.¹¹³ proposed a novel approach to understanding tamponade that lends itself to mechanical heart-lung interactions. Hyperinflation directly alters biventricular filling, whereas inspiration primarily alters RV filling and only indirectly affects LV volumes through changes in diastolic compliance. Thus, hyperinflation—as occurs in severe asthma and with the use of excessive amounts of PEEP—would produce a clinical picture indistinguishable from tamponade. Indeed, Rebuck and Read¹¹⁴ made this same observation in their analysis of the hemodynamic effects of severe asthma over 30 years ago, although they did not postulate a specific mechanism to explain this phenomenon. Presumably, the shift from “uncoupled” to “coupled” cardiac fossal restraint would occur as absolute lung volume increased, biventricular volume increased, or both. If cardiac volumes are small, and lung inflation does not overdistend the chest, RV filling will be primarily impeded. In contrast, in CHF states and with marked lung overdistention, both RV and LV filling may be compromised by ventilation.

Systemic Venous Return

Blood flows back to the heart from the periphery through low-pressure, low-resistance venous conduits. Guyton et al. characterized venous flow from the venous reservoirs into the right atrium.¹¹⁵ As downstream right atrial pressure varies, as occurs with ventilation, the rate of venous return inversely changes. Pressure in the upstream venous reservoirs is called mean systemic pressure. Mean systemic pressure is a function of blood volume, peripheral vasomotor tone, and the distribution of blood within the vasculature.¹¹⁶ Mean systemic pressure does not change rapidly during the ventilatory cycle, whereas right atrial pressure does owing to concomitant changes in ITP. Accordingly, variations in right atrial pressure represent the major factor determining the fluctuation in pressure gradient for systemic venous return during ventilation.^77,117 Positive-pressure inspiration increases ITP and right atrial pressure, decreasing the pressure gradient for venous return and RV filling,⁷⁹ and consequently, RV stroke volume.^{77,79,118–125} These physiologic effects have recently been validated in humans, using minimally invasive echocardiographic techniques wherein vena caval flow varies with the phase of the ventilatory cycle^25,126 (Figure 47-7). During normal spontaneous inspiration, the converse occurs: with decreases in ITP, right atrial pressure decreases, accelerating venous blood flow and increasing RV filling and RV stroke volume* (Figure 47-8).

Figure 47-7 Effect of increasing levels of CPAP on the relations between increasing airway pressure and right atrial pressure (left graph), airway pressure and intraabdominal pressure (center graph), and airway pressure and changes in RV end-diastolic volume in 43 postoperative fluid-resuscitated cardiac surgery patients.

(Derived from data in Van den Berg P, Jansen JRC, Pinsky MR. The effect of positive-pressure inspiration on venous return in volume loaded post-operative surgical patients. J Appl Physiol 2002;92[3]:1223-1231.)

Figure 47-8 Schematic representation of the effects of increasing or decreasing intrathoracic pressure (ITP) on steady state venous return. Note that decreases in ITP which decrease right atrial pressure to below zero relative to atmospheric pressure will only increase venous return by a limited amount, whereas increases in ITP will progressively decrease venous return to a complete circulatory standstill.

The decrease in venous return during positive-pressure ventilation is often lower than one might expect based on the increase in right atrial pressure. Fessler et al.¹²⁹ and Takata and Robotham¹³⁰ demonstrated in dogs that PEEP increases intraabdominal pressure by causing the diaphragm to descend, thereby increasing the pressure surrounding the intraabdominal vasculature. Because a large proportion of venous blood is in the abdomen, the net effect of PEEP is to increase mean systemic pressure and right atrial pressure. Accordingly, the pressure gradient for venous return may not be reduced by PEEP, especially in patients with hypervolemia. In fact, abdominal pressurization by diaphragmatic descent may be the major mechanism by which the decrease in venous return is minimized during positive-pressure ventilation.^131–135 Furthermore, Matuschak et al.¹³¹ found that although PEEP decreased blood flow to the liver in proportion to the induced decrease in cardiac output in normovolemic dogs, the liver’s ability to clear hepatocytic-specific compounds, such as indocyanine green, was unaltered. Finally, when cardiac output is restored to pre-PEEP levels by fluid resuscitation^131,136 while PEEP is maintained, liver clearance mechanisms increase above pre-PEEP levels.^136–139 These data are consistent with a PEEP-induced alteration in intrahepatic blood flow distribution. Thus, ventilation may have less of an effect on venous return than originally postulated, but the effect may be more complicated than we have imagined. Van den Berg et al.¹⁴⁰ examined the effects of varying levels of CPAP on right atrial pressure, intraabdominal pressure, and cardiac output in 42 postoperative cardiac surgery patients. Up to 20 cm H₂O CPAP did not significantly decrease cardiac output, as measured 30 seconds into an inspiratory hold maneuver. The reason for this apparent paradoxical effect became obvious when they compared the associated changes in right atrial pressure, abdominal pressure, and RV end-diastolic volume (Figure 47-9). What they documented was that only 30% of the increased airway pressure was transmitted to the right atrium. Perhaps more importantly, most of the increase in right atrial pressure was also realized by an increase in intraabdominal pressure, so it was not surprising that RV end-diastolic volume fell by less than 8% from pre-CPAP values. These data demonstrate that in the fluid-resuscitated patient, institution of positive-pressure ventilation may not result in a decrease in blood flow. However, if intraabdominal pressure is allowed to decrease, as would occur with an open laparotomy and decompression of tense ascites, a marked preload-responsive effect of positive-pressure ventilation can occur.

Figure 47-9 Echocardiographic and pulse Doppler images of superior vena caval flow patterns during positive-pressure ventilation. Note the inspiratory phase-dependent decrease in venous flow.

(Reproduced with permission from Jardin F, Vieillard-Baron A. Right ventricular function and positive-pressure ventilation in clinical practice: from hemodynamic subsets to respirator settings. Intensive Care Med. 2003;29[9]:1426-1434.)

With exaggerated swings in ITP, as occur with obstructed inspiratory efforts, venous return behaves as if abdominal pressure is additive to mean systemic pressure in defining total venous blood flow.^141–144 Recent interest in inverse ratio ventilation has raised questions as to its hemodynamic effect because its application includes a large component of hyperinflation. However, Mang et al.¹⁴⁵ demonstrated in an animal model of ALI that if total PEEP (intrinsic PEEP plus extra extrinsic PEEP) was similar, no hemodynamic difference between conventional ventilation and inverse ratio ventilation was seen.

Right Ventricular Filling

Under normal conditions, it is difficult to document any relation between RV filling pressure and volume. When RV filling pressure, defined as right atrial pressure minus Ppc, was directly measured in patients undergoing open chest operations, RV filling pressure was unaltered by acute volume loading.¹⁴⁶ Although right atrial pressure increased, Ppc also increased such that RV filling pressure remained unchanged. Similar data were seen when RV volumes are reduced by the application of PEEP in postoperative cardiac patients.¹⁴⁷ These findings suggest that under normal conditions, RV diastolic compliance is very high, and most of the increase in right atrial pressure seen during volume loading reflects pericardial compliance and cardiac fossa stiffness, more than changes in RV distending pressure. These data also imply that with RV filling, right heart sarcomere length remains constant. Presumably, conformational changes in the RV more than wall stretch are responsible for RV enlargement.¹⁵ Accordingly, changes in right atrial pressure do not follow changes in RV end-diastolic volume. When cardiac contractility is reduced and intravascular volume is expanded, RV filling pressure increases as a result of decreased RV diastolic compliance, increased pericardial compliance, increased end-diastolic volume, or a combination of all three. In support of this hypothesis, RV filling pressure does not increase until RV volume exceeds a certain threshold value.¹⁰⁵ Furthermore, in dogs with acute ventricular failure, volume loading increases Ppc more than ITP, consistent with pericardial rather than cardiac fossal restraint. If PEEP is increased in this setting, ITP but not Ppc selectively increases until ITP equals Ppc, then both ITP and Ppc increase equally if PEEP is increased further.⁷ In postoperative cardiac surgery patients,^13,22,148 PEEP—and by extension lung expansion—compresses the heart within the cardiac fossa in a fashion analogous to pericardial tamponade.

Venous return is the primary determinant of cardiac output.¹¹⁶ Since right atrial pressure is the backpressure to venous return, venous return is maintained near maximal levels at rest^51,125,126 because RV filling occurs with minimal changes in filling pressure.¹⁴⁸ The closer right atrial pressure remains to zero relative to atmospheric pressure, the greater the pressure gradient for systemic venous blood flow.^115,121 For this mechanism to operate efficiently, RV output must equal venous return, otherwise sustained increases in venous blood flow would overdistend the RV, increasing right atrial pressure. Fortunately, under normal conditions of spontaneous ventilation, the increase in venous return is in phase with inspiration, decreasing again during expiration as ITP increases.⁷⁷ Likewise, the pulmonary arterial inflow circuit is highly compliant and can accept large increases in RV stroke volume without changing pressure.^79,83 Thus, any increase in venous return is proportionally delivered to the pulmonary circuit without forcing the RV to increase its force of contraction or myocardial oxygen demand. Importantly, this compensatory system will rapidly become dysfunctional if RV diastolic compliance decreases or if right atrial pressure increases independent of changes in RV end-diastolic volume. An example of decreased RV diastolic compliance is acute RV dilation or cor pulmonale (pulmonary embolism, hyperinflation, and RV infarction) that induce profound decreases in cardiac output not responsive to fluid resuscitation. Dissociation between right atrial pressure and RV end-diastolic volume occurs during either tamponade or positive-pressure ventilation, because right atrial pressure is artificially increased by the increasing ITP. Accordingly, positive-pressure ventilation impairs normal circulatory adaptive processes. Furthermore, even if one restores the coupling of right atrial pressure and RV volume by using partial ventilatory support modes of ventilation, cardiac output will increase only if the RV can transduce the associated increase in venous return to forward blood flow. Thus, during weaning from mechanical ventilation, occult RV failure may be exposed and will manifest as a rapid rise in right atrial pressure and a fall in cardiac output. Since the primary effect of any form of ventilation on cardiovascular function in normal subjects is to alter RV preload via altering venous blood flow, the detrimental effect of positive-pressure ventilation on cardiac output can be minimized by either fluid resuscitation to increase mean systemic pressure^{4,118,140,141} or by keeping both mean ITP and swings in lung volume as low as possible. Accordingly, prolonging expiratory time, decreasing tidal volume, and avoiding PEEP all minimize the decrease in systemic venous return to the RV.^{1,21,77,120–124,149}

Since spontaneous inspiratory efforts increase lung volume by decreasing ITP, one sees an increase in venous return with spontaneous inspiration owing to the fall in right atrial pressure.^{19,67,121–123} However, this augmentation of venous return is limited^142,143 because if ITP decreases below atmospheric pressure, venous return becomes flow-limited as the large systemic veins collapse as they enter the thorax.¹¹⁵ This flow limitation is a safety valve for the heart because ITP can decrease greatly with obstructive inspiratory efforts,⁵² and if not flow-limited, the RV could become overdistended and fail¹⁵⁰ (see Figure 47-8). Still, in patients with decreased RV compliance, negative swings in ITP can augment RV filling. Interestingly, negative pressure ventilation, by augmenting venous return, was shown to increase cardiac output by 39% in intubated children following repair of tetralogy of Fallot.¹⁵¹ In this condition, impaired RV filling secondary to RV hypertrophy and reduced RV chamber size are the primary factors limiting cardiac output.

Left Ventricular Preload and Ventricular Interdependence

Changes in venous return must eventually result in directionally similar changes in LV preload, because the two ventricles are linked in series. For example, during a Valsalva maneuver, initially RV filling is reduced, but LV filling is unaltered.¹⁴ Then, as the strain is sustained, LV filling and cardiac output both begin to decrease.^108,152 This phase delay in changes from the RV to the LV is exaggerated if tidal volume or respiratory rate are increased and in the setting of hypovolemia.* Independent of this series interaction, direct ventricular interdependence can also occur and be clinically significant. Increasing RV volume shifts the intraventricular septum into the LV and simultaneously decreases LV diastolic compliance. During positive-pressure ventilation, RV volumes are usually decreased, minimizing ventricular interdependence.^{105,155–158} Echocardiographic studies document that although PEEP results in some degree of right-to-left intraventricular septal shift, the shift is small.^69,70 In fact, increases in lung volume during positive-pressure ventilation primarily compress the two ventricles into each other, decreasing biventricular volumes.¹⁵⁹ The decrease in cardiac output commonly seen during PEEP is due to a decrease in LV end-diastolic volume; in this setting, both LV end-diastolic volume and cardiac output are restored by fluid resuscitation^160,161 without any measurable change in LV diastolic compliance.¹⁰⁶

During spontaneous inspiration, RV volumes increase transiently, shifting the intraventricular septum into the LV,¹⁰⁶ decreasing LV diastolic compliance and LV end-diastolic volume.^103,158,162 This transient RV dilation–induced septal shift is the primary cause of inspiration-associated decreases in arterial pulse pressure; if greater than 10 mm Hg or 10% of the mean pulse pressure, the phenomenon is referred to as pulsus paradoxus.¹⁹ Since spontaneous inspiratory efforts can occur during positive-pressure ventilation and especially during partial ventilatory assist modalities, pulsus paradoxus can also be seen in mechanically ventilated patients.

Left Ventricular Afterload

Left ventricular afterload can be equated to systolic wall tension which, by the Laplace equation, is proportional to the product of transmural LV pressure and the radius of curvature of the LV, which itself is proportional to LV volume. Maximal LV wall tension normally occurs at the end of isometric contraction, reflecting both a maximal product of the LV radius of curvature (end-diastolic volume) and aortic pressure (diastolic pressure). Under normal conditions during LV ejection, LV afterload progressively decreases because LV volumes decrease markedly despite the small increase in ejection pressure. Importantly, when LV dilation exists, as in CHF, maximal LV wall stress occurs during LV ejection, since the maximal product of these two variables occurs at this time. Accordingly, LV afterload varies on the baseline level of cardiac contractility, arterial pressure, and intravascular volume. LV ejection pressure is the transmural LV systolic pressure, which can be approximated as transmural arterial pressure. Since normal baroreceptor mechanisms located in the carotid body tend to maintain arterial pressure constant with respect to atmosphere, if arterial pressure were to remain constant as ITP increased, LV wall tension would decrease as well. Similarly, if transmural arterial pressure were to remain constant as ITP increased, but LV end-diastolic volume decreased because of the increased ITP-induced decrease in systemic venous return, LV wall tension would also decrease.¹⁶³ Thus, by either mechanism, increases in ITP decrease LV afterload. Similarly, decreases in ITP with a constant arterial pressure will increase LV transmural pressure, increasing LV afterload.^14,164

Any process associated with marked decreases in ITP must also be associated with increased LV afterload and myocardial O₂ consumption (MVO₂). In fact, decreasing ITP, a common occurrence during spontaneous inspiratory efforts with bronchospasm and obstructive breathing, reflects an important cardiac stress inducing CHF. Furthermore, since weaning from positive-pressure ventilation to spontaneous ventilation may reflect dramatic changes in ITP swings and the energy requirements of the respiratory muscles, weaning from mechanical ventilation is a cardiovascular stress test.^163,165,166 Interestingly, Jabran et al.³⁷ demonstrated that all ventilator-dependent patients increased their cardiac outputs during weaning, but in those that failed to be liberated from mechanical ventilation, SvO₂ decreased, consistent with cardiovascular compromise. A similar argument can be given for the observed improvement in LV systolic function in patients with severe LV failure when placed on mechanical ventilation.¹⁶⁶ Interestingly, similar “auto-EPAP” effects of expiratory grunting have been reported in infants during crying⁷¹ and in an adult with severe LV failure.¹⁶⁷ It may well be that our failure to define specific physiologic parameters to describe subjects who may successfully be liberated from mechanical ventilation is because we do not include in these analyses measures of cardiovascular reserve.

Pulsus paradoxus occurs during spontaneous inspiration under conditions of marked pericardial restraint. This may occur because of pericardial limitations such as tamponade or constrictive pericarditis, as well as during loaded spontaneous ventilatory efforts when RV volumes swell and ITP decreases. In both cases, LV stroke volume decreases.^168–172 Perhaps the most prominent mechanism creating an inspiratory decrease in both LV stroke volume and systolic arterial pressure is the increased venous return–induced transient decrease in LV diastolic compliance that then results in decreased LV end-diastolic volume. The negative swings in ITP also increase LV ejection pressure (LV pressure minus ITP), increasing LV end-systolic volume.¹⁴ Other influences on LV systolic function can also occur during loaded inspiratory efforts, such as with obstructive sleep apnea. These include an increase in aortic input impedance,¹⁷³ altered synchrony of contraction of the global LV myocardium,¹⁷⁴ and hypoxemia-induced decreased contractility.¹⁷⁵ Hypoxia has the added detriment of also directly reducing LV diastolic compliance as well as decreasing myocardial contractile function.¹⁷⁶

Experimental repetitive periodic airway obstructions induce pulmonary edema in normal animals.^75,76 Furthermore, removing the negative swings in ITP by applying nasal CPAP results in improved global LV performance in patients with combined obstructive sleep apnea and CHF.¹⁷⁶

If ITP were to increase rapidly, as during a cough, arterial pressure would also increase by a similar amount such that both arterial pressure relative to ITP (transmural arterial pressure or LV ejection pressure)^14,177 and aortic blood flow¹⁰⁸ would remain constant. However, sustained increases in ITP must eventually decrease aortic blood flow and arterial pressure as a result of the associated decrease in venous return.¹⁴ Since normal baroreceptor-based homeostatic mechanisms tend to sustain a constant arterial pressure so as to maintain organ perfusion constant,⁵¹ if ITP increased arterial pressure without changing transmural arterial pressure, the periphery would reflexively vasodilate to maintain a constant extrathoracic arterial pressure-flow relation.¹⁵³ Since coronary perfusion pressure reflects the intrathoracic pressure gradient for blood flow, it is not increased by ITP-induced increases in arterial pressure. However, compression of the coronaries by the expanding lungs may obstruct coronary blood flow. Thus, the combined decrease in coronary blood flow may induce myocardial ischemia.^178–180

The effect of removing large negative levels of ITP is not similar to adding positive ITP on venous return. Relative increases in ITP from very negative values to zero, relative to atmosphere, will minimally alter venous return, whereas increases in ITP above atmosphere will impede venous return by increasing right atrial pressure. Once right atrial pressure becomes negative, venous return becomes flow limited. Very negative swings in ITP, as seen during vigorous inspiratory efforts in the setting of airway obstruction (asthma, upper airway obstruction, vocal cord paralysis) or stiff lungs (interstitial lung disease, pulmonary edema, and ALI), will selectively increase LV afterload. Such spontaneous inspiratory efforts may be the cause of LV failure and pulmonary edema that is often seen in these conditions,^18,52,75,76 especially if LV systolic function is already compromised^30,181 (see Figure 47-8). Similarly, removing large negative swings in ITP by either bypassing upper airway obstruction (endotracheal intubation) or through the institution of mechanical ventilation or PEEP-induced loss of spontaneous inspiratory efforts should selectively reduce LV afterload without significantly decreasing either venous return or cardiac output.* Reversing this argument, weaning from mechanical ventilation, with its associated increase in both metabolic demand and LV afterload, is a form of cardiac stress testing.

Using Changes in Intrathoracic Pressure to Define Cardiovascular Performance

Renewed interest in using the cardiovascular response to either spontaneous or positive-pressure ventilation as a means to diagnose cardiovascular responsiveness has recently occurred. However, sustained increases in airway pressure can also be used to measure cardiac contractility, as defined by the end-systolic pressure-volume relation (ESPVR).¹⁸³ Traditionally, LV ESPVR is generated by rapidly reducing LV preload by transient inferior vena caval occlusion. In the preload-dependent patient, passive inspiratory hold maneuvers that increase airway pressure by 5 to 10 cm H₂O (e.g., CPAP) will selectively decrease venous return without substantially altering either pulmonary vascular resistance or LV afterload, because the changes in lung volume and ITP are relatively small. Using this approach, Haney et al.^184,185 generated LV ESPVR similar to that created by transient inferior vena caval occlusion in a canine model. Since it is much easier to give 5 to 10 cm H₂O CPAP to patients than to pass an intravascular balloon into the inferior vena cava to induce transient vascular obstruction, this technique has promise as a means of selectively altering venous return in the bedside assessment of LV contractility. Denault et al., using the same logic, documented in mechanically ventilated patients during cardiac surgery that LV performance could be measured during ventilation using combined estimates of LV volume and ejection pressure.¹⁸⁶ As shown in Figure 47-10, ventilation induces profound dynamic changes in LV volumes consistent with rapidly varying changes in LV filling.

Figure 47-10 Effect of positive-pressure ventilation on LV volumes and related hemodynamic measures in a perioperative intact patient.

(Reproduced with permission from Denault AY, Gasior TA, Gorcsan J III, Mandarino WA, Deneault LG, Pinsky MR. Determinants of aortic pressure variation during positive-pressure ventilation in man. Chest 1999;116[1]:176-186.)

Clinical Application of Heart-Lung Interaction Physiologic Concepts: Predicting Preload Responsiveness

One of the fundamental uses of hemodynamic monitoring is to access intravascular volume status and predict the cardiac output response to volume loading.¹⁸⁷ Regrettably, measures of ventricular filling pressure and volumes, though useful in assessing cardiac performance, are very poor predictors of volume responsiveness.¹⁸⁸ Traditionally, the bedside clinician simply gave a rapid infusion of volume (e.g., 500 mL saline over 15 minutes) and noted whether cardiac output increased or not. Although effective at defining volume responsiveness, a volume challenge is time consuming and may delay appropriate treatment if there is no increase in cardiac output in response to this fluid challenge. Importantly, fully 50% of all hemodynamically unstable patients are not volume responsive¹⁸⁸; thus giving volume to all unstable patients will not benefit half. Consequently, interest in using the responses of the cardiovascular system to ventilation as a surrogate for a transient volume challenge emerged. Based on the previously described physiologic principles of heart-lung interaction, we know that positive-pressure inspiration will decrease venous return to the right ventricle, which will decrease filling of the left ventricle in 2 to 3 beats, resulting in a decrease in LV stroke volume if both ventricles are volume responsive. Under such conditions, LV stroke volume will decrease from apneic baseline values proportional to the IPPV-induced decrease in venous return and the slope of the LV function curve. Since arterial pulse pressure (diastolic to systolic arterial pressure) changes directly with LV stroke volume, one can use either stroke volume or pulse pressure variation to define volume responsiveness. Analysis of the arterial pressure waveform and its phase relationship to inspiration yield powerful tools to understanding the stroke volume and pulse pressure variations with inspiration. These variations are the basis for the prediction of preload responsiveness, and thus the application of the previously described physiologic considerations to the practical management of ventilator-dependent patients. A pulse pressure variation more than 13% or a stroke volume variation more than 10% on positive-pressure ventilation is highly predictive of volume responsiveness.¹⁸⁹

One of the most central aspects of cardiovascular homeostasis is the preload-dependent nature of LV performance. Documenting that LV EDV is above some minimal value, despite cardiac output and stroke work both being depressed, is essential for the diagnosis of cardiac pump dysfunction. Similarly, demonstrating that LV EDV is decreasing in the setting of hemodynamic instability presumes the diagnosis of inadequate circulating blood volume as the most likely cause of the hemodynamic instability, even though other etiologies (e.g., tamponade, cor pulmonale, restrictive cardiomyopathies) can coexist and require different treatments. However, knowing absolute LV EDV does not predict whether LV stroke volume will increase in response to volume loading.¹⁸⁸ Knowing the preload of a patient is not the same as knowing if the patient will be preload responsive. In addition, being preload responsive is not necessarily indicative of fluid requirement. Just knowing that a patient is volume responsive does not equate to the need for fluid resuscitation. One must also presume or document that inadequate blood flow exists, because normal subjects who are otherwise healthy may also be volume responsive but do not need fluid resuscitation.

Three techniques presently exist for defining preload responsiveness: (1) the traditional volume challenge, (2) noting the magnitude of the arterial pulse pressure or LV stroke volume variation during fixed tidal volume positive-pressure ventilation, and (3) determining the change in mean cardiac output in response to a passive leg-raising maneuver. The second of these three techniques is central to the clinical use of heart-lung interactions to diagnose and treat cardiovascular insufficiency.

Assessing Fluid Responsiveness During Positive-Pressure Ventilation

For either pulse pressure variation (PPV, ratio of maximal minus minimal pulse pressure to mean pulse pressure over 5 or more breaths) or stroke volume variation (SVV, ratio of maximal minus minimal stroke volume to mean stroke volume over 5 or more breaths) to reflect preload responsiveness, the tidal volume must be fixed, the sequential R-R intervals must be constant (i.e., no arrhythmias), and both parameters should be measured during unassisted positive-pressure breathing with a tidal volume (VT) of 8 mL/kg or more. Several studies have been conducted to assess the reliability of this technique in the assessment of volume responsiveness. Marik et al. summarized the published data in the last 10 years in a systematic review of the literature specifically “to determine the ability of dynamic changes in arterial waveform-derived variables to predict fluid responsiveness and compare these with static indices of fluid responsiveness.”¹⁹⁰ Twenty-nine studies, which enrolled 685 patients, were selected. The authors concluded that PPV, SVV, and systolic pressure variation (SPV) consistently predicted increments in stroke volume with fluid challenges. They also reported that the threshold values for such prediction of volume responsiveness were remarkably consistent throughout the studies and pointed to being between 11% and 13%. The diagnostic accuracy of these variables, as judged by area under the curve, was excellent (PPV 0.94, SPV 0.86, and SVV 0.84), with PPV significantly better than either SVV or SPV (P < 0.001). The efficacy of using either SVV or PPV metrics to predict fluid responsiveness has been validated in different clinical scenarios and populations, such as cardiac surgery,¹⁹¹ orthotopic liver transplantation,¹⁹² sepsis,¹⁹³ ARDS,^194–197 and critically ill children.¹⁹⁸

Effect of Tidal Volume on Pulse Pressure Variation and Stroke Volume Variation

Changes in tidal volume also alter the phasic swings in ITP and thus the dynamic changes in venous return. The greater the tidal volume, the greater the cycle-specific changes in venous return augmenting PPV and SVV at the same volume status. This is not surprising given that PPV is generated by the pressure transmitted from the airways to the pleural and pericardial spaces and thus will theoretically be decreased with low tidal volume ventilation and low pulmonary compliance. These interactions were reported by three separate groups of investigators. DeBacker et al.,¹⁹⁹ in a simple and elegant clinical description of the impact of changing tidal volume on PPV, showed that PPV varied directly with tidal volume. Renner et al.²⁰⁰ demonstrated that SVV also varied directly with tidal volume. Finally, Kim and Pinsky²⁰¹ showed that in anesthetized ventilated dogs, PPV decreased from 20.1 ± 10.8 to 9.5 ± 5.4% when tidal volume was decreased from 20 to 5 mL/kg. This can be conceptually described as the change in intrathoracic blood volume on a beat-to-beat basis as the difference between SV_RV and SV_IV associated with increasing tidal volume during intermittent positive-pressure ventilation. The greater the tidal volume the greater dynamic change in intrathoracic blood volume (Figure 47-11).

Figure 47-11 Effect of different tidal volumes (Vt) on the dynamic changes in intrathoracic blood volume (ITBV) as calculated from the differences in paired right ventricular stroke volume (SV_RV) to left ventricular stroke volume (SV_LV) for a single breath.

(Data from Mesquida, Kim, Pinsky in abstract form at the National Spanish Critical Care Medicine, 2010.)

PPV is generated by the pressure transmitted from the airways to the pleural and pericardial spaces, and thus will theoretically be decreased with low tidal volume ventilation and low pulmonary compliance. In addition, Muller et al.²⁰² showed that PPV failed to predict fluid responsiveness (defined as an increase of stroke index > 15%) after a fluid challenge (normal saline or hydroxy-ethyl-starch 6% 500 mL) in patients with low airway driving pressure (PPlat − PEEP < 20 cm H₂O). In that study, PPV over 13% or even over 7% was highly predictive of fluid responsiveness, but fluid responsiveness could not be ruled out in patients with lower PPV values.

Vistisen et al. have suggested that PPV should be indexed to tidal volume.²⁰³ In their study, they showed that PPV increased significantly with increments in tidal volume in three different states: hypovolemia, normovolemia, and hypervolemia. This approach is appealing because as Romand et al. previously showed,⁸ it is the change in lung volume, not in airway pressure, that determines the ITP change during positive-pressure breathing, and changes in ITP create the changes in venous return.

Effect of Positive End-Expiratory Pressure on Pulse Pressure Variation and Stroke Volume Variation

Since the primary effect of PEEP is to distend the lungs and increase ITP, PEEP normally reduces venous return and creates a functional hypovolemic state. Accordingly, Kubitz et al.²⁰⁴ showed in a porcine model that increasing PEEP levels increased both PPV and SVV. Interestingly, they also saw that this effect persisted in an open chest condition, albeit to a lesser degree. Potentially one could use the emergence of an increasing PPV during positive-pressure ventilation as a sign that lung recruitment has occurred and may have started to create hyperinflation.

Limits on the Ability of Arterial Pulse Contour to Assess Stroke Volume Variation

DeCastro et al.^205,206 validated the finding that stroke volume calculated by the PiCCO arterial pulse contour technique closely tracks steady state arterial pressure but does not track dynamic SVV. These data agree with a canine study by Gunn et al.²⁰⁷ that examined how accurately this same device tracked SVV as vasomotor tone was pharmacologically varied. These findings are important because they illustrate that the technique and device used to estimate SVV will have limitations minimizing its clinical utility. To reduce this inherent measurement bias across monitoring devices, clinicians should use only one minimally invasive device for the continuous measuring of SVV in the same patient over time.

Fluid Responsiveness Assessment During Spontaneous Ventilation

In spontaneously breathing patients and those with arrhythmias, the mean increase in flow 20 seconds after passive leg raising to 30 to 45 degrees gives a predictive value similar to PPV/SVV during positive-pressure ventilation. However, in this setting, the variable to measure is not PPV/SVV but the change in flow of greater than 10%. Passive leg raising (PLR) is a suitable test, given that it induces a gravitational transfer of blood from the periphery to the central circulatory compartment sufficient to significantly modify the preload of the LV and cause a change in flow or cardiac output.²⁰⁸

The best hemodynamic marker of passive leg raising as a predictor of volume responsiveness is a change in stroke volume, given that the test is transient in nature. Monnet et al.²⁰⁹ demonstrated that a change in descending aortic blood flow (measured by esophageal Doppler) more than 10% after passive leg raising was predictive of volume responsiveness in spontaneously breathing patients, with or without arrhythmia, and in sedated patients with sinus rhythm. Furthermore, Lamia et al.,²¹⁰ using transthoracic echocardiography, studied the stroke volume response to passive leg raising in patients breathing spontaneously. They showed that an increment in stroke volume over 12.5% was predictive of fluid responsiveness, with a sensitivity of 77% and a specificity of 100%.