A Deep Dive Into Retroesophageal Major Aortopulmonary Collateral Arteries

Introduction

Pulmonary atresia with ventricular septal defect and major aortopulmonary collateral arteries (PA/VSD/MAPCAs) is a relatively uncommon and complex form of congenital heart defect. ¹ The anatomy of MAPCAs can be highly variable, both in the number of MAPCAs to each lung and the anatomic origin and course of the MAPCAs. ² This anatomic variability becomes clinically relevant when MAPCAs are used in the reconstruction of a pulmonary vascular bed.^3–9 Many groups have adopted the “unifocalization” of MAPCAs for the treatment of PA/VSD/MAPCAs.^10–13

One anatomic variation of MAPCAs is those that traverse behind the esophagus en route to the lung parenchyma. The first report of retroesophageal collateral arteries dates back to an autopsy study published in 1948. ¹⁴ However, until recently, there was a paucity of information regarding the anatomy, physiology, and surgical characteristics of “retroesophageal” MAPCAs (or REMs). Our group at Stanford University has conducted a series of studies focusing on the subject of REMs over the past several years. This manuscript will synthesize these studies in a “deep dive into REMs.”

Anatomy of REMs

MAPCAs derive their embryologic origin from what will eventually become bronchial circulation. Early in fetal development, pulmonary blood flow is entirely derived from these bronchial vessels.^15–17 Once alternative sources of pulmonary blood flow are formed (ie, pulmonary artery and/or ductus arteriosus), the bronchial vessels typically regress and account for just 1% of cardiac output. ¹⁸ However, if these alternative sources of pulmonary blood flow do not materialize, then the bronchial vessels persist and maintain their status as the sole source of pulmonary flow (ie, MAPCAs). The majority of MAPCAs originate from the descending thoracic aorta and course anterior to the esophagus and airway, whereas REMs by definition course behind the esophagus.

REMs are always found on the side contralateral to the aortic arch (Figure 1). This is because the esophagus is always located on the opposite side of the arch. Specifically, patients with a left aortic arch will have a right-sided esophagus and thus can only have REMS to the right lung. Conversely, patients with a right aortic arch have a left-sided esophagus and thus can only have REMs to the left lung. This fundamental principle is helpful in evaluating both the preoperative pulmonary angiograms and intraoperative anatomy for REMs.

OPEN IN VIEWER

Figure 1.

Artist's illustration demonstrates the relationship between the intra-thoracic aorta and esophagus. In patients with a left aortic arch, the esophagus lies to the right of the aorta. Conversely, in patients with a right aortic arch, the esophagus lies to the left of the aorta. Because REMs pass behind the esophagus, they can only occur on the side contralateral to the aorta (From Mainwaring et al,¹⁹ with permission).

The anatomy of REMs differs in many ways from anteriorly located MAPCAs. One of these differences is that the origin of REMs is from the lateral surface of the descending aorta, whereas anterior MAPCAs originate from the anterior surface of the aorta or less commonly from the brachiocephalic vessels. When an REM is identified intraoperatively, we request that the anesthesiologist completely removes the transesophageal echocardiography probe in order to facilitate the dissection. The dissection is then carried out on both sides of the esophagus until the two dissection fields meet behind the esophagus. Some REMs pass through a portion of the esophageal muscle (ie, an intraesophageal course) requiring careful division of these slips of muscle to free the REM. REMs often have stenoses located at the point where it passes behind (or through) the esophageal muscle. This can be considered the hallmark angiographic sign of a REM (Figure 2). It should be noted that the majority of REMs will continue the posterior course and pass behind the bronchus before entering the lung parenchyma.

OPEN IN VIEWER

Figure 2.

Angiogram demonstrating a severe stenosis of an REM as it passes behind the esophagus. This is the hallmark or “sine qua non” of REMs (From Mainwaring et al,¹⁹ with permission).

In a study of 68 consecutive patients undergoing unifocalization procedures, 45 (67%) patients had REMs. ¹⁹ However, the prevalence of REMs was different depending on whether there was a left or right arch. For left aortic arches, 77% of patients had a REM to the right lung. In contrast, for right aortic arches, only 53% of patients had a REM that supplied the left lung. We had initially thought that patients could only have one REM. However, over the course of our study, we identified four patients (6%) with two REMs. Three of these patients had a left aortic arch while one had a right aortic arch.

Midsegment stenosis was present in 84% of REMs where the MAPCA crossed behind the esophagus. The degree of stenosis was much greater when the REM was intraesophageal than when it was not (80% vs 42%). One of the implications of these midsegment stenoses is that the REMs cannot simply be harvested and unifocalized. Instead, a decision must be made to either cut off the REM beyond the point of stenosis (assuming there is still adequate length) or patch the REM if the entire length is needed for unifocalization.

Another anatomic feature of REMs is the finding that they are associated with a higher prevalence of absent central pulmonary arteries compared with patients without REM (40% vs 20%). This two-fold disparity may be related to a “competition for space” in the central hilum during embryologic development. On this same theme, patients with REMs also have a higher prevalence of retroaortic innominate vein (10%).

To summarize the findings of this study, REMs were identified in two-thirds of patients but were more common in patients with a left arch than a right arch (77% vs 53%). REMs are only found on the side contralateral to the aortic arch, and the majority of REMs have significant stenosis located at the point where they pass behind the esophagus.

Physiology of REMs

As noted above, the course of REMs is behind both the esophagus and the bronchus. The net result is that these two structures separate REMs from the native pulmonary arteries in an anterior/posterior plane. Based on this observation, we hypothesized that REMs would be less likely to have connections with the native pulmonary arteries compared with their anterior MAPCA counterparts. ²⁰

To test this hypothesis, we reevaluated the preoperative cardiac catheterization images and matched these to the intraoperative observations and eventual decision to unifocalize or ligate any given MAPCA. MAPCAs were divided into three classifications, namely (1) single supply, meaning there was no connection to the pulmonary artery system (see video 1), (2) dual supply, but with an inadequate connection to the pulmonary arteries (see video 2), or (3) dual supply with an adequate connection to the pulmonary arteries (see video 3).

There are important implications for these three physiologic classifications of MAPCAs. Single-supply MAPCAs are the only source of pulmonary blood flow to an area of the lung, and therefore, must be unifocalized in order to have antegrade flow. Dual-supply MAPCAs do have a connection to the native pulmonary arteries. However, if the connection to the pulmonary arteries is smaller than the distal bed supplied by the MAPCA, this connection would prove restrictive if the proximal MAPCAs were simply ligated. Therefore, it is our recommendation to inspect (and sometimes probe) the connections between MAPCAs and pulmonary arteries and to unifocalize those connections that are restrictive. Finally, dual-supply MAPCAs with a large connection are those that can be safely ligated without the need for unifocalization.

Our study group for this study included 42 patients who underwent surgery at a median age of six months. For the 42 patients, there were a total of 187 MAPCAs or 4.5 MAPCAs per patient. Forty-eight MAPCAs were retroesophageal (six patients had two REMs) including 40 that were single supply, six were dual supply with inadequate connection, and two were dual supply with adequate connection. Based on this anatomy and physiology, 46 of the 48 (96%) REMs were unifocalized during surgery.

For the 139 anterior MAPCAs, 89 were single supply, 15 were dual supply with inadequate connection, and 35 were dual supply with adequate connection (Figure 3). Based on this anatomy and physiology, 104 of the 139 (75%) anterior MAPCAs were unifocalized during surgery. A comparison of REMs versus anterior MAPCAs demonstrated a significant difference in the percentage of MAPCAs unifocalized (96% vs 75%, P < .01).

OPEN IN VIEWER

Figure 3.

Bar graph demonstrating the percentage of single-supply (83% vs 64%, P < .01), dual supply with inadequate connection (13% vs 11%), and dual supply with adequate connection (4% vs 25%, P < .01) comparing retroesophageal versus anterior major aortopulmonary collateral arteries (From Mainwaring et al,²⁰ with permission).

To summarize this section, REMs had a significantly higher prevalence of being categorized as single-supply compared with anterior MAPCAs (83% vs 64%). It is likely that this difference is related to the extreme posterior course of REMs. These MAPCAs not only travel behind the esophagus but also by virtue of their position and trajectory, usually continue behind the bronchus as well. When the REMs eventually emerge posterior and lateral to the border of the bronchus, they are already at the edge of the lung parenchyma. As a consequence, REMs have little opportunity to connect with the native pulmonary artery system. This is confirmed by the data which showed that just 17% of REMs were categorized as dual-supply collaterals with the majority having an inadequate connection. The clinical takeaway is that nearly every REM (96%) will require surgical unifocalization.

Mid-Term Fate of Unifocalized MAPCAs

Given the extensive amount of information that we had garnered on this cohort of patients with REMs, we decided to evaluate the mid-term fate of the unifocalized MAPCAs. ²¹ Essential to performing this study was a comprehensive roadmap of the MAPCAs prior to intervention, operative details regarding which MAPCAs were unifocalized and which were ligated, and follow-up angiographic evaluation of the unifocalized pulmonary bed.

The study included 44 patients with a total of 197 MAPCAs. At surgery, 154 (78%) of the MAPCAs were unifocalized, whereas 43 (22%) were ligated because they were dual supply. Thirty-seven of the 44 patients subsequently underwent cardiac catheterization. The interval between unifocalization and catheterization is shown in Figure 4. Of the 154 unifocalized MAPCAs, 129 had angiographic follow-up, whereas 25 had no follow-up. For the 129 MAPCAs with follow-up, 123 (95%) had antegrade flow and 6 (5%) were occluded. Finally, of the 123 with antegrade flow, 97 (75%) were widely patent and 26 (20%) were stenotic (Figure 5).

OPEN IN VIEWER

Figure 4.

Figure demonstrating the percentage of patients on the y-axis and interval between the unifocalization procedure and follow-up catheterization procedure on the x-axis. The median interval was 17 months (From Mainwaring et al,²¹ with permission).

OPEN IN VIEWER

Figure 5.

Flow chart showing the fate of the unifocalized major aortopulmonary collateral arteries (MAPCAs). Of the 129 unifocalized MAPCAs with follow-up, 95% were found to be patent while 5% were occluded (From Mainwaring et al,²¹ with permission).

Thirteen of the 44 (29%) patients have subsequently undergone reintervention on a unifocalized MAPCA.^22,23 This includes nine who had balloon angioplasty, eight who had surgery, and four who had both. The Kaplan-Meier curve for freedom from reintervention is shown in Figure 6. Seven of the 13 patients had mild pulmonary vessel obstruction defined physiologically as no increase in pulmonary artery pressure. Four patients had moderate obstruction with peak systolic pulmonary artery to aortic pressure ratios between 0.45 and 0.60, and two patients had severe obstruction with pressure ratios greater than 0.60.

OPEN IN VIEWER

Figure 6.

Kaplan-Meier curve demonstrating freedom from reintervention. At three years, freedom from reintervention was 76% (From Mainwaring et al,²¹ with permission).

Figure 7 summarizes the results of reinterventions on unifocalized MAPCAs that were subsequently found to have stenoses. Of the five patients who underwent balloon angioplasty only, all five were mild cases and all five showed resolution. Of the four patients who underwent surgery only, two were mild cases and two were moderate and all four showed resolution.

OPEN IN VIEWER

Figure 7.

Flow diagram summarizing the type and result of reinterventions on unifocalized major aortopulmonary collateral arteries (MAPCAs) that were subsequently determined to have stenosis. Eleven of the 13 patients had angiographic resolution of the stenosis, while two patients do not have available follow-up (From Mainwaring et al,²¹ with permission).

Finally, for the four patients who underwent both balloon angioplasty and surgery, two cases were moderate and two cases were severe, and two of these cases demonstrated resolution, while two had no follow-up evaluation.

Literature Comparison

In the only other study in the literature that evaluated mid-term fate of unifocalized MAPCAs, d’Udekem et al published their experience from Melbourne, Australia in 2005 titled “Pulmonary atresia with ventricular septal defect and major aortopulmonary collateral arteries: Unifocalization brings no long-term benefits.” ²⁴ This study summarized data on patients who underwent a thoracotomy-based approach between 1975 and 1995. Thus, these data are from many decades ago and utilized an entirely different surgical approach than is currently in use.

There were 31 patients in this study who had follow-up angiography with a total of 60 MAPCAs evaluated (ie, two MAPCAs per patient). The results of this study demonstrated that 26 of the 60 (43%) unifocalized MAPCAs were occluded and 12 additional MAPCAs (20%) were stenotic. Thus, only 22 of the 60 (37%) unifocalized MAPCAs were widely patent. Given these very low patency rates, it is not surprising that most patients did not derive any clinical benefit from these procedures. It was this observation that led Dr Jane Sommerville to call for a moratorium on thoracotomy-based unifocalization procedures in the early 1990s.^25,26

One of the unfortunate consequences of the poor results achieved with a thoracotomy-based approach with MAPCAs was that many surgical groups worldwide gravitated to a nonunifocalization approach, or what has been called “pulmonary artery rehabilitation.”^27,28 This approach relies on the placement of central shunts or right ventricle to pulmonary artery conduits to increase pulmonary artery flow and pressure with the idea that this will promote the growth of the pulmonary arteries and ultimately a satisfactory pulmonary vascular bed. However, there are several important flaws in this concept. First, the majority of patients with PA/VSD/MAPCAs do not have normal arborization of the pulmonary arteries (ie, the pulmonary arteries do not go to all 18 segments of the lung). This means that the pulmonary artery rehabilitation approach will only recruit those segments of the lung that are in continuity with the pulmonary arteries. The remaining segments of the lung derive their source of blood flow from single-supply MAPCAs, which by definition do not communicate with the pulmonary arteries and thus these segments will be excluded from the pulmonary vascular bed if the MAPCAs are not unifocalized. The number of lung segments supplied by single-supply MAPCAs varies considerably from patient to patient, but it is safe to say that the more segments of lung excluded from the pulmonary vascular bed the higher the pulmonary vascular resistance and consequently the higher the postoperative right ventricular pressures will be if the VSD is surgically closed. Yet another misnomer regarding pulmonary artery rehabilitation is the idea that the addition of increased flow and pressure will be uniformly transmitted throughout the pulmonary bed and result in uniform growth. In fact, the entrance site of MAPCAs into the pulmonary arteries is typically at right angles, and the blood flow is channeled in both directions. This is the same physiologic situation that is responsible for creating coarctation of the aorta, and indeed, the exact same phenomenon occurs with MAPCAs and pulmonary arteries. Thus, if one establishes a proximal source of pulmonary blood flow (eg, shunt) and ligates the MAPCA at its origin, there almost invariably will be stenosis in the pulmonary artery at the entrance of the MAPCA (observation of the authors).

These stenoses prevent the transmission of pressure and flow to the distal pulmonary vascular bed and thus inhibit the growth of that bed as well. For these multiple reasons, it appears that the Achilles heel of the pulmonary artery rehabilitation approach is high postoperative right ventricular pressures.

Conclusions

To summarize, REMs are anatomically and physiologically different than their anterior counterparts. Because of their posterior location, the possibility of interconnections to the pulmonary arteries is markedly diminished. The clinical implication is that nearly every REM will require unifocalization. Follow-up evaluation of unifocalized MAPCAs performed via midline unifocalization revealed a 95% patency rate. These results account for the excellent clinical results that have been achieved using the midline unifocalization approach.

Supplemental Material