Is there any role for an atrial septal defect in off-loading a borderline left ventricle?

Abstract

A small atrial septal defect with right-to-left shunt is useful for off-loading a dysfunctional right ventricle postoperatively. However, an atrial septal defect with left-to-right shunt may not be as useful for a dysfunctional left ventricle. Experimental data are limited at present. Thus, we reconsider the related physiology to guide future approach.

INTRODUCTION

In paediatric cardiac surgery, a small atrial septal defect (ASD) is often maintained/created in patients at risk of postoperative right ventricular (RV) dysfunction because a dysfunctional RV may struggle to provide sufficient pulmonary blood flow (Qp) for adequate left ventricular (LV) filling (preload), which is essential for sustaining systemic blood flow (Qs). An ASD, by allowing right-to-left shunting, improves LV preload and therefore Qs, although at the expense of Qp. This is tolerable provided Qp is not diminished excessively so that oxygen saturations remain acceptable; a small/moderate drop in saturations may be tolerated thanks to a better Qs. Nevertheless, this strategy is mainly for cyanotic patients (typically with Tetralogy of Fallot), who are adapted to lower saturations. This strategy also off-loads the right heart, but this is a secondary benefit; the primary objective is to secure Qs.

However, a small ASD is also sometimes advocated in biventricular repair in neonates and young infants at risk of postoperative LV dysfunction (typically due to borderline size) and consequent high left atrial pressure (LAP) [1]. This strategy, unlike that for RV dysfunction, is not proposed for sustaining Qs but for reducing LAP by left-to-right shunting to avoid pulmonary oedema. This is a much more complex concept with little data available to support or refute it but does deserve a discussion of arguments both ways based on cardiovascular physiology.

ARGUMENTS AGAINST THIS STRATEGY

Cardiovascular physiology functions in a manner that sustains Qs as a matter of utmost priority, which depends closely on maintaining many factors including adequate volemia, LV compliance to allow the necessary LV preload and contractility to provide output. LV preload, however, also depends on an adequate LAP; the lower LV compliance is, the higher LAP must be to secure adequate preload. Indeed, LAP reflects LV compliance (diastolic function) in the sense that, in the absence of mitral valvar stenosis, LAP matches the LV end-diastolic pressure (LVEDP) at all levels of volemia [2, 3]. Therefore, the best way to reduce LAP is to improve LV compliance rather than to reduce LAP directly.

LV compliance (diastolic function) is determined by intrinsic myocardial factors but is also affected by afterload. Compliance decreases as afterload increases because a higher afterload requires stronger myocardial contractions to sustain output, which increases wall stress and, as a reaction, LV thickness increases and LV volume decreases, thus tending to decrease wall stress (the law of Laplace). Such reduction in compliance may also be explained in terms of the energetics of actin–myosin cross-bridge formation, which are such that the stronger the myocardial contraction is, the harder its relaxation: LVEDP and LAP increase [4]. This is why hypertensive crises may cause pulmonary oedema. Independently of aetiology, however, a dysfunctional LV is less compliant than normal (LVEDP and LAP are higher), because diastole is the first process that suffers when a ventricle has problems [4].

LAP, high or low, simply indicates the state of LV compliance as reflected in LVEDP; the only way to reduce LAP and maintain adequate Qs is to improve LV compliance. Even left-to-right shunting through an ASD by itself may not reduce LAP unless LV compliance is also somehow improved, otherwise LAP would not decrease because of an auto-regulatory feedback loop; the shunt augments RV preload and therefore output, which increases LA filling by the same amount that was shunted [2, 5]. In other words, the LA is not off-loaded because all volume that is diverted from it returns to it immediately by anatomic and physiologic imposition. Conversely, this feedback loop would also prevent augmentation of LAP in response to ASD closure because elimination of the shunt has the opposite effect of the above—it decreases RV preload and output, and therefore LA filling, by the same amount.

Deliberate reduction of LAP without improving LV compliance would also reduce LV preload and output. This would be counter-productive, since these babies have a borderline LV with an output already at the lower limit of the normal range. A lower output causes sympathetic activation, thus aggravating matters with tachycardia (reducing filling time) and vasoconstriction (increasing afterload). Also, left-to-right shunting through an ASD causes volume overload of the RV and pulmonary circulation [2].

Such auto-regulatory feedback loop also applies to the right atrium (RA) given that the cardiovascular system is a closed loop; all blood shunted away from either atrium returns to that atrium. This raises the question of how an ASD with right-to-left shunting can off-load the RA (although off-loading the RA is not the objective; sustaining Qs is)?

How right-to-left shunting through an atrial septal defect off-loads the right heart

The atria may be off-loaded in only 2 ways:

1. By reducing the circulating volume with diuretics (cautiously: this also reduces Qs).

2. By improving the compliance of their respective ‘downstream’ ventricle.

The second mechanism is the basis of how an ASD may off-load the RA. If the RA’s usual corresponding downstream ventricle (the RV) is dysfunctional, and therefore its compliance is reduced, an ASD diverts part of RA blood towards the LV instead, which is also downstream to the RA and is healthy with a normal compliance (i.e. it may be preloaded adequately without generating a high LVEDP), thus off-loading the right heart. The dysfunctional RV is partially bypassed, and the LV is used as RA’s downstream ventricle; the shunted blood takes a lower-resistance short-cut forward towards the systemic circulation. This reduces Qp, which is well tolerated as long as the reduction in oxygen saturation is not excessive. This process does not trigger the sympathetic nervous system because Qs is maintained.

Conversely, the LA cannot be off-loaded in this manner because the systemic ventricle cannot be bypassed as there is no other ‘downstream’ ventricle that could offer an alternative ‘forward’ path, i.e. towards the systemic circulation. The key is that both ventricles are downstream to the RA, but only the LV is downstream to the LA in a biventricular circulation. All left-to-right-shunted blood enters the RV (which is upstream to the LA) and therefore returns to the LA without passing through the systemic circulation.

Lack of data

Data proving the above arguments are limited. Nevertheless, excellent results have been achieved without an ASD in biventricular repairs of hearts with borderline LVs [6, 7]. Additionally, examples of maintenance of LAP are noted, although in older patients with normal ventricles: when closing ASDs percutaneously, LAP does not increase at all (Fig. 1, published with permission of the ethics’ committee of the state of Andalusia, Spain, reference 1326-N-22, waiving the necessity for written consent). Furthermore, on personal observation (A.-R.H.) in the setting of biventricular repair in 2 neonates with aortic arch hypoplasia, borderline LV and small ASD, ASD closure was done after termination of cardiopulmonary bypass with an over-and-over suture that had been placed loosely on the ASD earlier and passed out of the heart through the RA roof. Again, LAP was unaffected by ASD closure and the suture was tied. However, an ASD is useful in certain exceptional circumstances, which must be recognized.

The exceptional situations where an atrial septal defect is useful

In ‘hybrid circulations’

In neonates with a small/borderline LV, the RV can be used to support the systemic circulation. Such a ‘hybrid circulation’ is established with the Norwood operation or bilateral pulmonary arterial banding and ductal stenting. This requires an ASD to direct part of the pulmonary venous return to the RV. Anatomic repair may be possible later if the LV grows well; this may be boosted by making the ASD restrictive to encourage LV preload (so-called LV rehabilitation strategy). In this situation, the ASD does off-load the LA because the RV is reorganized into being ‘downstream’ to the LA and is healthy, offering an alternative forward path (with lower resistance) to the systemic circulation. The dysfunctional LV is bypassed, and the RV takes over the function of sustaining Qs. Given that Qs is sustained, sympathetic reflexes are not triggered.

Preoperatively in 2 circumstances

1. In neonates with critical stenosis/hypoplasia of a left-sided structure, the RV must secure Qs via the arterial duct while awaiting surgery. This is like the above-mentioned ‘hybrid circulation’, which requires an ASD.

2. In infants with significant LV volume overload due to post-tricuspid left-to-right shunting, an ASD partially off-loads the LV while awaiting surgery. This is because shunting depends on 3 factors: (i) the size of the defect, (ii) the difference in pressures between the two sides and (iii) the difference in systemic and pulmonary vascular resistances. However, when the defect is large enough to allow unrestricted flow across, the volume of the shunt becomes dependent purely on pressures and resistances, i.e. larger defects (or additional defects) do not cause more shunting. In the absence of an ASD, the entire shunt volume passes through the LV, i.e. the full burden of volume overload is on the LV. An ASD allows some of that shunting to occur without passing through the LV, although this causes some RV volume overload. Total left-to-right shunting is the same, since this is determined solely by vascular resistances in such malformations, but the burden of volume overload is shared between the two ventricles, with less LV dilatation and lower LVEDP and LAP while awaiting repair [8].

In patients on extracorporeal membrane oxygenator support secondary to left ventricular dysfunction

Here, an ASD off-loads the LA and LV with no risk of reducing Qs because this is sustained by an extracorporeal pump.

In a specific situation of biventricular dysfunction

In patients with LV diastolic dysfunction secondary to chronic RV dilatation/dysfunction caused by an ASD, closure of this defect may worsen LV function, with elevation of LVEDP and LAP. This phenomenon has not been convincingly explained yet but is thought to be because chronic RV dilatation makes this ventricle dependent on a high preload for adequate functioning while the LV is chronically compressed and may become dependent on a volume-overloaded RV to splint the septum [9]. The ASD helps by maintaining RV volume overload, thus avoiding a deterioration in LV function. This is typically seen in older patients (often adults) with a large ASD and chronic mitral regurgitation. The acute change in loading conditions on the LV can cause acute pulmonary oedema, and leaving a fenestrated ASD may allow the LA to decompress in the early post-operative period while the circulation re-adjusts, although only at the cost of reduced Qs.

ARGUMENTS IN FAVOUR OF THIS STRATEGY

The arguments in favour of this strategy are clinical observations; many surgeons have seen patients with intact atrial septum and high LAP after biventricular repair, in whom creating an ASD allowed recovery. Also, in a study of repair of total anomalous pulmonary venous connection in neonates and young infants, opening the vertical vein (physiologically equivalent to creating an ASD) in patients who were haemodynamically unstable reduced LAP and pulmonary arterial pressure while systemic blood pressure increased suggesting increased cardiac output [10]. These observations may possibly be explained in terms of the Starling curve; the LV in these patients may be on the flat or descending parts of the curve, where a reduction in LV preload and LAP would cause no change or even an augmentation in cardiac output.

Lastly, an ASD is beneficial because it allows noninvasive (echocardiographic) monitoring of LAP. A very small ASD would be sufficient for this. Such a small ASD is unlikely to be haemodynamically significant (Qp/Qs <1.5).

COMMENTS

The strategy of using an ASD to ‘offload’ a dysfunctional LV is of questionable value except in some very specific situations; it should only be used after careful consideration of the physiolgy. This strategy really needs to be clarified with better and more objective data. In the meantime, perhaps a compromise should be done, namely leaving only a small ASD (<4 mm). This is likely sufficient for the supposed benefits (if these are real) yet small enough to avoid significant right-sided volume overload.

Data are difficult to obtain because of the rarity of these conditions, the ethics of studies on small babies and the variability in patients’ response to ASDs depending on where on the Starling curve their LV may be. However, with computational modelling and artificial intelligence, it may be possible to model patients’ circulation, simulate postoperative conditions and assess ASD management individually. This may be the way forward.

FUNDING

None declared.

Conflict of interest: none declared.

DATA AVAILABILITY

All data included in this work are totally anonymized and are contained within the article itself. There are no other data associated with this article.

REFERENCES