Unbalanced forms of atrioventricular septal defect (AVSD) continue to be challenging and present poor surgical outcomes, especially in those patients with mild or moderate hypoplasia of the left ventricle (LV), also known as “borderline” patients. Unbalance can strike different levels independently, as separate parts of a whole: the atria, the atrioventricular valve (AVV), the ventricular inflows and the ventricular cavities. Being “balanced” at one level does not necessarily mean that this condition will be fulfilled in the rest. Moreover, finding a certain degree of asymmetry at one level does not imply that the same degree will be found in the others.

Unbalance is related to a lack of symmetry in: a) size or b) distribution of blood flow between the pulmonary and systemic circulations. The first might also be described as “anatomic unbalance”, and reflects the actual absence of symmetry in size (diameter, length, volume) between the right and left components of the heart.

On the other hand, asymmetry in blood flow distribution can be conceived as a “hemodynamic unbalance”. Certain anatomic features such as the sizes of the atrial septal defect and the ventricular septal defect (VSD) or malalignment of the atrial and/or ventricular septums are more related to distribution of blood flow between the systemic and pulmonary circulations than to the presence or absence of symmetry in size between right and left structures (although they might contribute to the development of anatomic asymmetry).

The aim of this study is to validate a surgical decision-making algorithm based on a geometric model of interplay between different anatomic components of the crux of the heart evaluated by echocardiography. The final goal is to be able to identify the best surgical approach in terms of morbidity and mortality. Severity of anatomic unbalance must be assessed at all different levels to identify those cases with high risk of mortality of one-stage biventricular repair (BVR). Hemodynamic factors such as the amount and direction of the shunt between right and left structures must also be assessed, as they influence the ability of the LV to maintain the systemic circulation.

For many years, the modified atrioventricular valve index (mAVVI) has been used to characterize patients in balanced or unbalanced. It is defined as the ratio between the left and the total areas of ​​the AVV in a subcostal view. The same relationship is maintained by measuring diameters of the left component (LC) of the AVV and the total AVV in 4-chamber (4C) view (Figure 1). Balanced patients have a mAVVI between 0.4 and 0.6, while right dominant unbalanced ones show values below 0.4. Those cases with a mAVVI between 0.2 and 0.39 display mild or moderate degrees of unbalance (borderline cases), while those with mAVVI < 0.2 are considered severely unbalanced.

Figure 1: 4C view, end diastole. A: Total AVV annulus diameter (red); B: the VSD (white) is measured as the line extends from the tip of the interventricular septum and crosses the AVV perpendicularly; C: the LC of the AVV (black dots) is measured as the portion of the AVV extended from the left hingepoint of the AVV to the intersection of the VSD line.

Thus, if the LC represents more than 40% of the AVV, the case can be considered as balanced, whereas if it represents less than 20%, the case will show severe unbalance.

Likewise, it was described in previous publications that patients with a large VSD had higher risk of mortality when submitted to BVR as compared to those with smaller VSD. It is correct to assume that those patients with a small VSD have a LV that was already handling most of the systemic cardiac output in the preoperative state, since there was no significant left to right shunt through that small VSD. On the other hand, in patients with a large VSD, the LV is unloaded and its ability to maintain cardiac output after a BVR is at least uncertain. Despite all this, until now, no precise limits have been described regarding how large the VSD has to be to make the biventricular strategy prohibitive and lead us to choose univentricular repair (UVR) or alternative surgical strategies.

Years ago, one of the largest multicentric studies on AVSD reported on the behavior of the angle formed between both hingepoints of the AVV and the crest of the interventricular septum. This angle was called RV/LV inflow angle. The study revealed that patients with a more obtuse angle were more balanced while those with more acute angles had greater degree of unbalance.

Excellent results have been achieved with this technique in terms of mortality, morbidity and postoperative pathway patency. Connections between the pulmonary venous return and the left atrium are wide, and even in small children, anastomosis of 2.5 cm x 1 cm can be performed.

In a previous study, our group showed that the value of the angle is essentially an indirect way of assessing the size of the VSD relative to the degree of unbalance. The sides of this angle in conjunction with the AVV form a triangle in which the main angle is the RV/LV inflow angle and the height represents the size of the VSD (Figure 2).

Figure 2: L: left hingepoint of the AVV, R: right hingepoint of the AVV, A: crest of the interventricular septum, M: point of intersection between the VSD line and the AVV, AVVs: AVV diameter, VSDs: VSD diameter, LCs: diameter of the left component of the AVV, RCs: diameter of the right component of the AVV, RIs: right ventricular inflow, LPAs: left primary annulus, α: RV/LV inflow angle.

The value of the angle is much more affected by the size of the VSD than by the degree of unbalance. In other words, the degree of unbalance has a scarce impact on the final value of the angle. Conversely, there is a strong correlation between the size of the VSD and the value of the angle: the larger the VSD, the sharper the angle will be (Figure 3).

Figure 3: Triangles similar to those in the previous figure are compared. All share the same base, which represents the AVV. The two points located on the base represent the division between the right and left components of AVV, and therefore the mAVVI. The vertices A, B, C and D identify different locations of the crest of the interventricular septum. The height of each triangle then represents the size of the VSD. A and B have different mAVVI but the same size of the VSD. The same correlation is found between C and D. In both situations, the RV/LV inflow angle of the two triangles is similar. However, A and C, like B and D, share the same mAVVI respectively, but the sizes of the VSD differ significantly, determining a different value of the angle.

It is important to note that the size of the VSD is relevant only if it is evaluated relative to another structure that correlates with patient's size. For this purpose, we chose the AVV diameter, which is an anatomically adjacent structure with a diameter quite correlated with the size of the patient. A VSD of a certain diameter will not have the same hemodynamic impact in a small patient as compared to a larger one. Thus, the concept of the indexed ventricular septal defect (inVSD) emerges as a new echocardiographic variable obtained as the ratio between the VSD size and the total diameter of the AVV in apical 4C view (Figure 1).

With the aid of simple mathematical equations that describe the geometry of the set of triangles, the variables mAVVI, RV/LV inflow angle and inVSD can be correlated. The mAVVI represents the degree of unbalance, and the inVSD, the size of the VSD. The angle is useful to delimit different subgroups. Using the mean and standard deviations (SD) of the angle reported by the Congenital Heart Surgeons´ Society in a multicentric study, we created a graphic of inVSD versus mAVVI where the different values ​​of the angle are represented (Figure 4).

Figure 4: InVSD vs mAVVI graphic. Each curve represents a value of the angle. For values ​​of mAVVI between 0.2 and 0.39, the different curves are almost horizontal and parallel to each other, indicating that for each value of the angle a small range of inVSD values ​​corresponds. This means that the correlation between angle and inVSD is very strong. It can also be observed that the value of the angle decreases as the size of the inVSD increases. The highlighted curves represent the mean value of the angle (108 °) and the values ​​of -1 SD (81 °) and + 1 SD (135 °) for the unbalanced population studied by the Congenital Heart Surgeons' Society.

In our previous work, together with the description of the inVSD as a new echocardiographic variable, we proposed an algorithm of surgical decision-making based on this parameter. Within the population with mild and moderate degree of unbalance (mAVVI between 0.2 and 0.39), patients with low values ​​of inVSD will have better chances to successfully overcome BVR as compared to those patients with high values of this parameter, in whom univentricular palliation or alternative approaches might represent the best strategy (Figure 5).

Figure 5: Areas are delimited according to the mean and SD of the RV/LV inflow angle. For mAVVI < 0.2, univentricular approach is recommended due to severe unbalance. For mAVVI values ​​between 0.2 and 0.39, the value of the inVSD will suggest the surgical strategy to be followed. For low inVSD values, BVR can be considered as a safe strategy, while high values ​​of this parameter suggest that univentricular palliation or other alternative strategies need to be considered (UVR: univentricular repair, BVR: biventricular repair).

Even with the introduction of this new index in preoperative assessment of this very challenging population, we were facing an incomplete examination. As we know, malformations of the AVV and its subvalvar apparatus are frequent associations in uAVSD. When analyzing the potential ability of the LV to manage systemic cardiac output after septation, we cannot ignore that these associations can modify the "true inflow" to the ventricle located below. In other words, even achieving a good diameter of the mitral component of the AVV during BVR, the presence of subvalvar derangements (mitral parachute, fused papillary muscles, single papillary muscle and thickening of the chordae tendinae and presence of accessory cords) can cause obstruction at the true entrance to the LV and affect its performance.

Taking this into consideration, Szwast and colleagues described in 2011 the left ventricular inflow index (LVII) and found that patients with values under 0.55 had increased risk of mortality of BVR. This index is defined as the ratio between the secondary annulus (SA) and the left primary annulus (LPA), both measured in apical 4C view. The first is represented by the smallest diameter of color Doppler at LV inflow and is obtained by tracing a line between the crest of the interventricular septum and the LV wall (Figure 6).

Figure 6: Echocardiographic measurements: AVV: atrioventricular valve, LC: left component of the AVV, VSD: ventricular septal defect, SA: secondary annulus, mAVVI: modified atrioventricular valve index, inVSD: indexed ventricular septal defect, inSA: indexed secondary annulus.

The LPA is measured as a line extended from the left hingepoint of the AVV to the crest of the interventricular septum. In our geometric model, this line represents the hypotenuse of the left right-angled triangle (Figure 2). Its size depends on the size of the other two sides of the triangle, given by the VSD and the left component of the AVV. A large VSD determines a large LPA. Given the fact that the LVII is calculated as the SA / LPA ratio, an increase in LPA will determine a low value of LVII. This is why this index is not a "pure" measure of the inflow into the LV, since it is very influenced by the size of the VSD (Figure 7).

Figure 7: LVII: LV inflow index, VSD: ventricular septal defect, LC: left component of the AVV, LPA: left primary annulus. Keeping the value of the left component of the AVV fixed (and therefore the degree of unbalance), the LPA value will depend on the size of the VSD. Considering that the LVII is calculated as the ratio between LC and LPA, the higher the value of LPA, the lower the value of LVII will be. Therefore, a large size of the VSD will determine a low value of the LVII.

For this reason, the incorporation of a "pure" echocardiographic parameter for the evaluation of the mitral subvalvar area can provide important information for surgical decision-making. In our latest work, we introduced a new index called indexed secondary annulus (inSA). This parameter is defined as the ratio between the SA and the total diameter of the AVV. It represents the width of the true inflow to the LV in relation to the width of the AVV. By means of the geometric evaluation of the already described triangles and the SA, we developed a formula that relates the inSA with the mAVVI, the inVSD and the LVII.

Suppose that the left component of the AVV (LC) is equal to the true inflow to the LV represented by the SA, that is, there is no significant obstruction given by the left subvalvar apparatus. This means that inSA = mAVVI, since these indexes reflect the relationship of SA and LC relative to total AVV diameter, respectively. If ​​we introduce into our formulas these values and the cutoff value of LVII = 0.55 validated by Szwast, we will find that inVSD / mAVVI = 1.52. In the graphic of inVSD versus mAVVI, this limit is plotted as a dotted line for the mAVVI interval between 0.2 and 0.39 (Figure 8).

Figure 8: The dotted line represents the cutoff value of LVII = 0.55 for those cases in which the diameter of the left component of the AVV (LC) is equal to the diameter of the true inflow of the LV given by the secondary annulus (SA) (that is, there is no obstruction in the mitral subvalvar apparatus). Given that LVII = 0.55 represents the limit below which there is an increased risk of mortality of BVR, cases falling above this line will show an increased risk due to the large size of the VSD.

Thus, an inVSD / mAVVI ratio = 1.52 implies that the VSD is one and a half times larger than the size of the mitral component of the AVV (LC). This value then represents the limit above which the large size of the patient's VSD in relation to its degree of unbalance determines increased risk of mortality of BVR in a single stage strategy. In other words, biventricular correction in a single stage approach should be avoided in all patients with inVSD / mAVVI > 1.52, even if there is no significant obstruction at the mitral subvalvar apparatus.

The slope of the dotted line depicted in Figure 8 is not random: the lower the degree of unbalance, the larger the size of the VSD that will be tolerated. Conversely, a small VSD will be required to follow a biventricular track in cases displaying more severe degrees of unbalance.

So far, we have clearly established the cutoff value of the inVSD in the absence of any kind of obstruction determined by the left subvalvar apparatus. However, if there is some degree of obstruction at this level, the SA is smaller than the LC. This means that the amount of blood passing through the mitral component of the AVV is greater than the one that actually enters the LV. In this situation it is necessary to stablish whether the severity of the subvalvar obstruction is such that it determines an increased risk of mortality of BVR. For that, we can calculate the critical value of inSA (or "critical inSA") for each combination of mAVVI and inVSD by introducing the cutoff value of LVII = 0.55 into our equations. This critical value represents the inSA below which the subvalvar obstruction is significant enough to determine an increase in the risk of mortality of BVR (Table 1).

Table 1: Critical values ​​of inSA for given values ​​of mAVVI and inVSD.

Thus, knowing the mAVVI and the inVSD we can obtain the corresponding critical inSA. After measuring patient’s inSA, we can compare it with its critical value. If patient’s inSA is larger than the critical inSA, we can assume that there is no significant obstruction at the mitral subvalvar apparatus. On the other hand, if patient’s inSA is smaller than the critical value, important obstruction will be found at the true LV inflow level.

Finally, using these indexes we can refine surgical decision-making in this population (Figure 9).

Figure 9: Suggested algorithm for surgical decision-making in patients with right dominant unbalanced AVSD (BVR: biventricular repair, UVR: univentricular repair).

Initial stratification is made using mAVVI. Patients with mAVVI < 0.2 are not considered candidates for BVR due to severe hypoplasia of left-sided structures. Therefore, UVR is recommended.

The second level of stratification is used in cases of mild and moderate unbalance (mAVVI between 0.2 and 0.39) by means of the inVSD / mAVVI ratio. As explained before, this value represents the limit above which there is an increased risk of mortality of BVR given by a too large VSD, even without anatomic obstruction at the “true” LV inflow. As both indexes are related to the AVV, their ratio reflects the relation in size between the VSD and the mitral component of the AVV. A VSD one and a half times larger than the mitral component indicates elevated risk of mortality of one-stage BVR, suggesting that UVR or a staged approach must be considered. To choose between both strategies we can use the net value of inVSD. In our previous work, we found that an inVSD > 0.6 indicates that the VSD is "excessively large", implying that the most reasonable strategy could be univentricular palliation. This cutoff value obtained using the RV/LV inflow angle matches the cutoff value of the dotted line in Figure 8 for a mAVVI = 0.39.

For inVSD < 0.6 in this context, the suggested strategy is biventricular correction in stages as described by Foker, which attempts to induce growth of the hypoplastic LV by increasing blood flow through the AVV. First stage consists in partial closure of the VSD and the atrial septal defect, division of the AVV and pulmonary artery banding. Geometrically speaking, partial VSD closure is the surgical way to reduce the inVSD. As a part of this strategy, a portion of the right component of the AVV is often partitioned with the VSD patch onto the LV side. Geometrically speaking again, this procedure increases the size of the LC of the AVV, and therefore, the value of mAVVI. Then, combination of VSD size reduction and LC enlargement decreases inVSD / mAVVI ratio and prepares the patient for a second stage in which this ratio might have become more propitious for BVR.

Even in this staged approach, LV inflow derangements must be addressed as a third level of stratification, using inSA as compared to its critical value (Table 1). If patient´s inSA is above critical inSA, staged BVR repair is advisable. In the presence of LV inflow obstruction determined by inSA below its critical value, this strategy can be followed only if the subvalvar apparatus is amenable of repair. Among other surgical techniques, secondary chordal resection, accessory LV outflow tract tissue resection, chordal shortening and delamination or splitting of the papillary muscles might be used. If obstruction at that level cannot be relieved, UVR is suggested as the safest approach.

On the other hand, an inVSD / mAVVI ratio ≤ 1.52 means that the VSD is not too large to preclude one-stage BVR. Nevertheless, the inSA must be used as a third level of stratification to evaluate adequacy of “true” LV inflow. If patient´s inSA is above the critical inSA depicted in Table 1, BVR can be achieved, as the LV inflow is free from significant obstruction. But if patient´s inSA falls below the critical inSA, there is elevated risk of mortality of BVR and this approach will only be feasible if LV inflow obstruction is relieved at the time of the operation. If the subvalvar apparatus is not amenable of repair, UVR should be considered.

This stratification must also evaluate and take into consideration other relevant anatomic, physiologic and individual features, which include, but are not limited to, ductal dependence of the systemic circulation, AVV straddling, apex-forming LV, atrial septal malalignment and left and right ventricular volumes. Besides, trisomy 21, AVV regurgitation and pulmonary hypertension play an important role, as they are related to higher morbidity and mortality after UVR.