The human heart undergoes developmental changes throughout childhood that are important to consider in the management of PARDS (see Table 1). In utero, high PVR facilitates the redirection of systemic and placental venous return away from the lung through either the ductus arteriosus or the foramen ovale for eventual ejection to the systemic circulation. As the morphologic RV (assuming situs solitus with levocardia and D-looped ventricles) is conditioned by elevated PVR in utero, RV and left ventricle (LV) wall thicknesses are nearly identical at birth (16). As the PVR drops postnatally with a neonate’s first breaths, RV afterload rapidly declines. This permits the gradual reconditioning of the RV in a low pressure environment that results in thinning of the RV wall mass over the ensuing weeks-months of infancy.

At a cellular level, myocardial fibers of the neonatal and young infant’s heart globally have higher connective tissue-to-contractile protein ratios with generally fewer and less organized myofibrils present per cardiomyocyte (17). This myofibril anatomy compromises the ability of the cardiomyocyte to both contract and to relax, leading to a state of minimal systolic and diastolic reserve. The transverse tubules and sarcoplasmic reticula overlying myofibrils within cardiomyocytes are also immature in the young infant, further limiting the calcium-dependent inotropic capacity of the infant’s myocardium and rendering the myocardium reliant upon extracellular calcium sources for sarcomeric contraction. Moreover, the neonatal heart has a higher preponderance of parasympathetic innervation with lower β-adrenergic receptor expression compared to older children or adults, limiting RV and LV contractile reserve.

The density and anatomic arrangement of RV fibers also contribute to the limited contractile reserve of the RV. In contrast to the LV that is comprised of three myocardial fiber layers with complex alignment permitting torsional constriction of the LV cavity, the RV is generally made up of only superficial and deep layers of muscle fibers (18). The superficial myofibers are predominantly transversely oriented, blending into the superficial myocardial layer of the ventricular septum and LV, while the deeper myofibers are more longitudinally aligned (18). Given the lower postnatal wall stress perceived by the RV, each myocardial fiber within the RV carries a substantially reduced number of mitochondria (19). These factors together limit RV contractile capacity and metabolic reserve at baseline in both neonates and younger children.

In older children as in adults, the RV generally has a greater capability of tolerating sudden myocardial demands (e.g., increases in preload and/or afterload) than neonates through elevations in heart rate, contractility, and stretch to accommodate the increased RVEDV. Moreover, extrapolating from canine models of acute afterload changes to the ventricles, LV contractility may contribute 20%–40% of RV output in older children through ventricular tethering (20). In neonates, such a demand on the RV is not as well tolerated. In addition to the cellular and anatomic differences in the neonatal myocardium described above, the resting heart rate of neonates is typically higher than older children. Thus, the “therapeutic window” by which neonatal heart rates can elevate to generate a compensatory increase in cardiac output before tachycardia limits ventricular filling is narrower than older children. Additionally, the PVR in some infants may remain elevated for the first several months of life, increasing basal afterload on the RV. In children with persistently elevated PVR from birth as seen in bronchopulmonary dysplasia or congenital heart diseases with pulmonary over-circulation, the poorly compliant RV is able to gradually adapt to the higher levels of wall stress through myocardial hypertrophy. However, this compensatory mechanism comes at the cost of further limiting RV diastolic function (21).

Given the limited contractile reserve of the RV in an older child, acute increases in RVEDV are initially tolerated through modest dilation of the relatively compliant RV wall. As greater stress is placed on the RV myocardium through increasing afterload and/or preload and myofibrils are stretched further, myosin-actin interactions are reduced and systolic function becomes embarrassed (extreme of the Frank–Starling relationship). The reduction in RV contractility is thus unable to respond to the increased preload by raising, or even maintaining, stroke volume. The ensuing acute RV dilation can cause a precipitous compression of the LV through ventricular interdependence. In combination with diminished pulmonary venous return due to limited RV output and reduced pulmonary blood flow, LV compression can result in an embarrassment to systemic cardiac output—a process termed acute cor pulmonale (ACP). The combination of increased RV wall stress and decreased systemic cardiac output can result in coronary ischemia that further reduces RV systolic function that can culminate in cardiac arrest.

During PARDS, the developing heart of a neonate or young infant is rather suddenly exposed to a potentially toxic cardiorespiratory milieu. Intrathoracic positioning intrinsically subjects the right atrium and RV to transpulmonary pressures changes. During positive pressure ventilation with high mean airway pressures, right atrial filling is limited due to a reduction in transmural pressure, which may decrease RV preload (though this effect may be diminished as lung compliance worsens). While positive pressure ventilation generally reduces transmural pressures across both ventricles (which would be expected to reduce ventricular afterload and metabolic demand), increases in PVR observed during PARDS can overwhelm the modest reduction in RV afterload and stress the RV myocardium. As PARDS severity worsens, PVR rises due to hypoxia-/hypercarbia-mediated vasoconstriction, reduction in recruited alveolar units, and higher mean airway pressures generated in an effort to sustain systemic oxygenation. These stressors to the RV are not easily correctable. The RV of an older child generally has the capacity to handle acute increases in RVEDP and RVEDV by mounting an increase in RV contractility and moderate dilatation (Figure 2). As discussed above, the neonatal myocardium has much less ability to respond to acute changes in volume and pressure experienced during severe PARDS.

To the credit of the PALICC investigators, the RV is mentioned as a potential culprit worthy of interrogation in the setting of “suspected cardiac dysfunction” during PARDS (10). Unfortunately, absence of a formal definition for RV dysfunction in children precludes uniform diagnosis or management. Even in adult practice, precise definitions for RV dysfunction and RV failure remain elusive. These pathophysiologic manifestations may be more practically distinguished by the RV myocardial response to increased RVEDV, as offered by Vieillard-Baron et al. (23) (Figure 2). In states of RV dysfunction, though myocardial contractility may be impaired, the ventricle is able mount a response to higher preload by increasing, or at least maintaining, stroke volume without developing systemic venous congestion. As RV systolic function deteriorates, the RV is unable to increase or maintain stroke volume in response to incremental increases in RVEDV, leading to impaired RV outflow. In this state of RV failure, reduced RV output cascades into lower LV preload; systemic-to-suprasystemic RVEDP with resultant RV ballooning, LV compression, and tricuspid regurgitation; and ultimately systemic venous congestion with reduced end-organ perfusion.

While a general gestalt exists amongst pediatric cardiologists and intensive care providers, an objective, operational definition of pediatric RV dysfunction or failure remains elusive for a number of important reasons. First, there is not a consensus among pediatric cardiologists regarding whether RV dysfunction should be categorized according to systolic dysfunction, diastolic dysfunction, or both. Second, while most pediatric cardiologists and pediatric intensive care providers would agree that a battery of biomarkers and echocardiographic data are essential to diagnosing RV dysfunction at the bedside, the precise assays and ultrasonic readouts required for diagnosis continue to be debated. Moreover, many echocardiographic measures commonly employed to determine RV function in adults remain unvalidated in children and are rarely performed by pediatric sonographers.

There is little evidence to guide the management of invasive ventilation when RV dysfunction is suspected or confirmed in a child with PARDS beyond PALICC recommendations. Informed by clinical data from ARDS studies in adults (3), it is reasonable to employ the following strategies to reduce PVR and RV afterload. As oxygen is a selective pulmonary vasodilator, F_IO₂ may be temporarily increased and mean airway pressure judiciously titrated to optimize dynamic compliance or respiratory system impedance (assuming negligible contribution of airway resistance to the patient’s respiratory mechanics) in an effort to preserve FRC (thus maximizing alveolar recruitment while limiting regional dead-space) and drive oxyhemoglobin saturation above 90%. Caution is warranted here as prolonged exposure to high F_IO₂ may contribute to ventilator-induced lung injury, and exposure to high mean airway pressure without re-evaluation of RV performance may precipitate further RV decline. Permissive hypercapnia goals should be tightened, though a precise upper limit of PaCO₂ is unclear at this time. Acidemia should be corrected either with judicious increases in minute ventilation (while limiting driving pressure) and/or with increased circulating levels of bicarbonate. There is insufficient evidence to support mode of ventilation (e.g., conventional mechanical ventilation, high frequency oscillatory ventilation, airway pressure-release ventilation, etc.) to accomplish these goals, regardless of RV function (54). At present, it is unknown whether spontaneous respiratory effort is more advantageous for a dysfunctional RV during PARDS than neuromuscular blockade.

Prone positioning as a treatment for ARDS and a tool to decrease ventilator-induced lung injury is well-described in adults (55–58). It has also been shown to have a role in unloading the RV and improving right ventriculo-arterial coupling (43). Proning can decrease PVR and increase cardiac output among adult patients with ARDS (44). Vieillard-Baron et al. (45) demonstrated an increase in cardiac index with 18 h of prone positioning of patients with ARDS-induced ACP. Moreover, an adult cohort with severe ARDS randomized to prone posture experienced a significantly lower incidence of cardiac arrest relative to those who remained supine (55). There are several proposed mechanisms for the hemodynamic benefit of prone positioning that culminate in reduced RV afterload (43). Despite increasing dorsal transpulmonary pressure (as measured by esophageal manometry) (59), the prone posture may increase overall respiratory system compliance by reducing the ventral-to-dorsal transpulmonary pressure differential (60). This improvement in respiratory mechanics reduces driving pressures while increasing the homogeneity of alveolar ventilation (43), limiting risk of ventilator-induced lung injury and associated RV dysfunction (61, 62). Improvements in ventilation and oxygenation associated with more optimal ventilation/perfusion matching would be expected to reduce PVR. Moreover, greater homogeneity in lung aeration would be expected to improve FRC toward baseline. Even modest restoration in FRC could reduce PVR and thus decrease RV afterload. There is also speculation that proning optimizes RV three-dimensional geometry, leading to improved RV systolic function (45). Though the evidence and mechanisms by which prone positioning promotes reductions in RV stress during ARDS have only been demonstrated in adults (summarized nicely by Vieillard-Baron et al. (43)), in the absence of evidence in the pediatric population, prone positioning can be reasonably employed for children with or at risk for acute RV dysfunction in the setting of PARDS.

Much like the goal of reducing systemic vascular resistance (SVR) in the setting of LV systolic failure, therapies that directly target PVR reduction are logical to prescribe in the setting of PARDS-related RV systolic dysfunction. The most commonly prescribed first-line treatment for increased PVR during PARDS is inhaled nitric oxide (iNO) (46) given its demonstrable reduction of PVR in adults with ARDS (63). Mechanistically, iNO induces pulmonary arterial vasodilation in regions of well-ventilated lung by stimulating guanylate cyclase in pulmonary arteriolar smooth muscle cells to generate cyclic guanosine monophosphate (64). However, despite evidence that iNO improves oxygenation during PARDS, children do not perceive a survival benefit with its use (65). Though other inhaled pulmonary vasodilators [e.g., inhaled epoprostenol (or prostacyclin) and iloprost (a synthetic analog of prostacyclin)] may have similar effects on PVR and systemic oxygenation as iNO, these therapies are intensely potent and may result in rebound pulmonary arterial hypertension if inadvertently discontinued. Therefore, their use in the PICU remains restricted to the management of isolated pulmonary hypertension without coincident PARDS (46–48).

Orally administered systemic vasodilators, such as sildenafil (phosphodiesterase 5 inhibitor), ambrisentan (endothelin receptor antagonist), bosentan (endothelin receptor antagonist), and riociguat (stimulator of guanylate cyclase), are typically utilized in children with chronically elevated PVR from diseases such as bronchopulmonary dysplasia, congenital heart disease, or sickle cell anemia (47, 66). In children with known pulmonary disease who present to the PICU with severe PARDS, critical care providers commonly continue these home therapies even in the absence of RV dysfunction for the hemodynamically stable child. Alternatively, in the previously healthy child now presenting with severe PARDS, our experience suggests that these therapies are not commonly considered unless the patient manifests persistent RV systolic dysfunction during a prolonged course of PARDS. In the face of a dearth of evidence to guide the use of inhaled or systemic vasodilators to treat or prevent RV systolic dysfunction during PARDS, these therapies cannot be universally recommended in all children with severe PARDS (10).

In the absence of robust evidence to support the use of inotropic or vasoactive agents for RV systolic dysfunction with PARDS, it is logical to reach for these therapies to sustain RV contractility and the LV contribution to RV output in an effort to maintain forward pulmonary blood flow and prevent RV bowing into the LV cavity. Low dose epinephrine (<0.05 μg/kg/min) or dobutamine (1–20 μg/kg/min) are catecholamines used to promote inotropy through β₁ G-protein coupled receptors with variable activity on pulmonary arterial vasodilatation through β₂-receptors (50, 67). The benefit of epinephrine and dobutamine over other inotropes (e.g., digoxin, milrinone) lies in their capacity for rapid titration. However, both catecholaminergic agents improve inotropy at the cost of increased myocardial oxygen demand.

Calcium is an important inotrope in an infant with RV dysfunction. Prior work has demonstrated that infants treated with intravenous calcium following cardiopulmonary bypass demonstrated significant improvements in cardiac output and mean systemic blood pressure compared to infants who did not receive calcium (68). Similarly, multiple case series attest to the incidence of myocardial dysfunction in infants manifesting nutrition-mediated hypocalcemia (69, 70). Physiologically, these findings may be explained by the underdeveloped t-tubular system of the myocardial sarcomere and an underdeveloped sarcoplasmic endoreticulum (evidence of which has been helpfully summarized by Baum and Palmisano (17)). The developmental immaturity of these myocardial structures demand that sarcomeric contractility in an infant’s heart relies heavily on extracellular ionized calcium availability to facilitate myosin-actin interactions. Therefore, ensuring normal circulating levels of ionized calcium in young children may be necessary to sustain RV systolic function during PARDS. In a similar way, it may be logical to leverage the calcium-sensitizing effects of levosimendan to sustain or improve RV systolic function in a child with PARDS. However, the quality of evidence for the use of levosimendan in children with primary cardiac disease remains poor (71), and the evidence for levosimendan use outside of acute-on-chronic heart failure in children or adults is wholly lacking.

Milrinone may be considered in a hemodynamically stable children RV dysfunction due to PARDS. Milrinone is a phosphodiesterase 3 inhibitor, promoting pulmonary and systemic vasodilation along with myocardial inotropy by decreasing the degradation of cyclic adenosine monophosphate within vascular smooth muscle and myocardium, respectively. However, milrinone should be used with caution. It may promote global pulmonary vasodilatation that could worsen the respiratory shunt fraction and exacerbate systemic hypoxemia. Moreover, the long half-life and renal clearance of milrinone may precipitously lead to refractory hypotension in a child developing impaired renal function (72). In a hemodynamically unstable patient, milrinone is best used in conjunction with a vasopressor if used at all (49). In the setting of ACP, decreased SVR caused by milrinone could theoretically reduce LV end-diastolic pressure and paradoxically worsen LV compression by the ballooning RV (46).

Norepinephrine has been suggested in experimental models to improve RV function and cardiac output (4). By increasing SVR through the activation of α₁-receptors, norepinephrine raises the systemic diastolic pressure and thus may improve coronary perfusion. Norepinephrine also has mild activity on myocardial β₁-receptors that can improve both RV contractility and the LV contribution to RV cardiac output. In addition, norepinephrine increases LV afterload that results in higher LV end-diastolic pressures to compete against the rightward-shifting interventricular septum during RV failure (73). However, as with the use of epinephrine, care must be taken to recognize that norepinephrine will increase myocardial oxygen demand in an already stressed heart. Vasopressin, on the other hand, may be used in the hemodynamically unstable child with PARDS-mediated RV systolic dysfunction to maintain systemic arterial pressures without directly increasing myocardial oxygen demand. Vasopressin raises SVR by activating V₁-receptors in systemic arterioles, thus promoting increased intracellular calcium availability through the activation of phospholipase C. Importantly, vasopressin has a smaller effect on PVR than SVR (74). The mechanism is thought to be due to V₁-receptor-mediated nitric oxide release in the pulmonary vasculature that leads to vasodilation (51). Vasopressin has been shown to consistently decrease the pulmonary-to-aortic systolic pressure ratio in pediatric patients with known pulmonary hypertension (50). These properties suggest vasopressin as an ideal vasopressor choice in the setting of hemodynamically unstable RV systolic failure (46).

Fluid management during PARDS is a complex task. The conservative approach of fluid restriction and/or diuresis may reduce extravascular lung water and thus improve ventilation/perfusion matching; however, this may come at the cost of reduced intravascular volume and end-organ perfusion. Current recommendations for fluid management in the setting of PARDS focus on goal-directed care (neither conservative nor liberal in approach) (75). Typically, a child with normal biventricular function handles the significant changes in intravascular volume that occur between initial volume loading during resuscitation of shock and the aggressive diuresis that commonly follows cardiovascular stabilization in the setting of PARDS. However, RV function is particularly sensitive to the complex cardiopulmonary changes that occur with intravascular fluid shifts, especially during positive pressure ventilation with high mean airway pressure. An accurate assessment of intra- and extravascular volume status of a child in acute RV dysfunction is thus critical for preserving and/or restoring RV performance during PARDS.

The notion of preload dependence in a failing RV has merit, and increasing intravascular volume in an acutely hemodynamically unstable patient may be necessary. However, excess preload can worsen RV dilatation, resulting in septal bowing into the LV, tricuspid regurgitation with worsening venous congestion, and increased RV myocardial wall tension that compromises coronary perfusion pressure and may precipitate clinical decompensation (18, 52, 76–78). Though pulmonary arterial catheters are used exceedingly rarely in pediatrics, invasive monitoring using CVP trends can be helpful to guide the need for decongestion and to better understand the right atrial pressure necessary to provide adequate preload (53). Decongestion allows for decompression of the RV, reducing ventricular interdependence, and improving hemodynamics overall (49). Decongestion is primarily achieved through judicious diuresis, typically with the use of intravenous loop diuretics (e.g., furosemide, bumetanide) with or without the use of thiazide diuretics (e.g., intravenous chlorothiazide or, in children with sustained gut function, metolazone). There are many barriers to effective diuresis in patients with right heart dysfunction including acute kidney injury mediated by high CVP, low cardiac output, and the resulting reduction in renal perfusion pressure (79). In a hypervolemic patient with acute RV dysfunction, vasopressors may be required to sustain sufficient renal perfusion pressure while diuretic therapy is use to achieve RV decompression and venous decongestion (49). For children who fail to respond to diuretic therapy, continuous renal replacement therapy (CRRT) may be necessary to achieve intravascular volume removal (80). However, the need for CRRT in the setting of PARDS-associated RV dysfunction should heighten the pediatric intensivist’s alertness to the patient’s manifestation of extremis and impending cardiovascular collapse. In such a clinical situation, the ethical considerations and risks for deploying CRRT must be strongly weighed against any perceived benefit to the patient.

The decision to deploy extracorporeal membrane oxygenation (ECMO) for severe PARDS refractory to lung-protective ventilation is often challenging and emotionally charged. The pediatric intensivist often has to guide a family through the complex risks and benefits of deploying ECMO for their child in a time-sensitive manner with limited information. When RV dysfunction develops in the setting of severe PARDS, complex treatment decisions can become substantially more intricate. Here we would like to discuss (1) the decision to use venovenous (VV) versus venoarterial (VA) ECMO as the initial cannulation strategy for PARDS complicated by RV dysfunction and (2) specific treatment decisions around RV dysfunction during VV ECMO.