Discussion

Our retrospective study of hearts with tetralogy of Fallot stored in historical archives has shown that the majority have two-leaflet pulmonary valves. Most commonly this occurs in the setting of a bisinuate root without leaflet fusion, with a smaller proportion possessing a functionally bileaflet valve housed within a trisinuate pulmonary root. When seeking to segregate the hearts using cluster analysis, we were able, with good statistical significance, to identify three basic patterns. The largest group was made up of hearts with bisinuate and bileaflet valves. The second group was made up of hearts with trisinuate and trifoliate valves, whilst the smallest group had functional bifoliate valves in a trisinuate root. Previous studies have shown that it is the rule for the hypoplastic pulmonary root in the setting of tetralogy of Fallot to be guarded by a valve with less than three functional leaflets. In one study, just under three-fifths of the cohort were found to have two-leaflet valves. ^{Reference Altrichter, Olson, Edwards, Puga and Danielson7} In another study, using intraoperative inspection, found almost nine-tenths of the roots to have only one or two leaflets. ^{Reference Liu, Jiang and Peng8} Neither of these studies, however, nor any other studies of which we are aware, sought to further clarify the morphology of the abnormal roots in terms of the number of valvar sinuses.

Our current study used specimens from historical archives, which made it impossible for us to make clinical correlations. In this regard, as far as we can establish, it is only the study using intraoperative inspection that has provided such correlations. ^{Reference Liu, Jiang and Peng8} These investigators found that those with other than trisinuate and trileaflet valves were more likely to have lower preoperative arterial saturations, smaller diameters of the virtual basal ring, and greater preoperative right ventricular outflow tract obstruction. The patients, furthermore, were less likely to get a valvar-sparing repair, and more likely to require pulmonary arterioplasty, left pulmonary arterioplasty, and valvar annuloplasty. They also tended to have a longer postoperative duration of mechanical ventilation, and longer postoperative stay in the intensive care unit. In longer-term follow-up, they were also more likely to require reintervention. These findings stress the ongoing need to establish precise valvar morphology during the preoperative work-up. Our findings now indicate that, in addition to the number of functional leaflets, attention is also required regarding the number of valvar sinuses. As we have emphasised, little attention seems to have been paid thus far to the number of sinuses within the hypoplastic pulmonary root.

We have previously demonstrated that, during its development, the outflow tract is best analysed as having distal, middle, and proximal parts. These give rise, respectively, to the intrapericardial arterial trunks, the arterial roots, and the ventricular outflow tracts. ^{Reference Anderson, Lamers and Hikspoors12,Reference Henderson, Eley and Chaudhry13} The components of the aortic and pulmonary valves, therefore, form within the middle part of the developing outflow tract. In the early stages of development, the outflow tract has walls that are exclusively myocardial, and the lumen is lined with cardiac jelly. It is with ongoing development that new non-myocardial cells are added to the outflow tract from the second heart field. With this addition of new material, there is effective proximal regression of the myocardial borders, with the distal myocardium then surrounding the developing arterial roots. At the same time, the initially common lumen is separated into the aortic and pulmonary channels by fusion of the outflow cushions, which extend in a spiralling fashion through the parts of the tract that have retained their myocardial walls. The cushions fuse from distal to proximal, eventually producing complete physical separation of the systemic and pulmonary channels. ^{Reference Anderson, Lamers and Hikspoors12} The distal surfaces of the fused major cushions also excavate so as to produce the valvar leaflets. These components interdigitate with similar excavating aortic and pulmonary intercalated swellings, thus producing two primordiums, each with a trileaflet pattern. Addition of non-myocardial tissues derived from the second heart field then produces the walls of the valvar sinuses. By then, the proximal major cushions have united and muscularised, building a shelf in the roof of the right ventricle so as to commit the aortic root to the developing left ventricle. Tubercles from the atrioventricular cushions then close the persisting aortic-right ventricular communication, producing what will become the membranous part of the septum. At this stage, both the pulmonary and aortic valvar leaflets remain supported entirely by myocardial tissue. It only after the completion of ventricular septation that the inner heart curve within the left ventricle undergoes fibrous conversion, thus producing fibrous continuity between the leaflets of the mitral and aortic valves. ^{Reference Anderson, Lamers and Hikspoors12}

On the basis of our knowledge of normal development as described above, we can now infer that tetralogy of Fallot, along with double outlet right ventricle with subaortic interventricular communication and pulmonary stenosis, is the result of incomplete transfer of the developing aortic root to the left ventricle. This is likely the consequence of an abnormal cranial location of the septal attachment of the proximal cushions, which at the same time narrows the subpulmonary outflow tract. This makes it impossible for the tubercles of the atrioventricular cushions to close the aorto-right ventricular communication. As we have also described, the pulmonary and aortic valves form similarly to one another. Their right and left leaflets, along with the septal leaflets of the atrioventricular valves, are derived from the endocardial cushions. Their nonadjacent leaflets, however, are derived from the intercalated valve swellings, sometimes referred to as the intercalated cushions. ^{Reference Henderson, Eley and Chaudhry13} Thus, the leaflets themselves of each valve are not derived from structures with the same developmental origin. Although the intercalated valve swellings eventually appear similar to the endocardial cushions, they are different at a cellular level. Despite these differences, both the leaflets derived from the cushions, and those formed from the intercalated valve swellings, enlarge into the lumen of the developing outflow tracts and ensure one-way flow in the separate outflow tracts. ^{Reference Nomura-Kitabayashi, Phoon and Kishigami14} Through flow-triggered mechanisms yet to be fully delimited, the endocardial cushions and the intercalated valve swellings become smaller, producing the leaflets in a complex process of valvar sculpting. This involves a process of excavation, with the leaflets never “elongating”, as suggested by current conventional wisdom. With the added information of the cellular mechanisms involved, ^{Reference Henderson, Eley and Chaudhry13,Reference Nomura-Kitabayashi, Phoon and Kishigami14} we can now make additional inference regarding the appearance of the different abnormalities we have observed. Presence of only two leaflets in the pulmonary root could imply failure of formation of the pulmonary intercalated valvar swelling. Such a mechanism would account for the commonest pattern we encountered, namely the bisinuate and bileaflet pulmonary root. The findings in the hearts with bileaflet but trisinuate roots imply fusion between the cushions during ongoing development, as would be the case for those having valves with only one leaflet and a central orifice. The varying orientation of the zone of apposition between the leaflets is likely to reflect the specific cushions undergoing fusion.

Clinically, the findings of this study can have clinical implications. This has been highlighted by the study by Liu and colleagues who demonstrated that the presence of an abnormal pulmonary valve was associated with lower preoperative arterial saturations and higher preoperative right ventricular outflow tract gradients. Abnormal pulmonary valves were also associated with increased likelihood of requiring a main pulmonary arterioplasty and left pulmonary arterioplasty. Those with abnormal pulmonary valves were also less likely to undergo a valve sparing repair and undergoing a transannular patch. The findings of the current study may also be associated with similar clinical approaches although this was not investigated in this study. Additional studies applying these criteria to children with tetralogy of Fallot undergoing surgical repair would be of clinical utility. Interest to the surgical technique associated with clusters as well as the future need for reintervention would be of clinical interest.

The features of interest included in this study are able to be assessed by echocardiography. Right ventricular outflow tract dimensions, pulmonary valve leaflet number, and pulmonary valve sinus number can all be assessed with echocardiography. CT can also demonstrate all these features with even greater detail.

We emphasise that our study has been made possible only by access to specimens from historical archives. The use of such specimens has permitted us to make an efficient and thorough evaluation of the overall morphology of the pulmonary root. Thus far, there is a paucity of published data regarding the overall characteristics of the pulmonary root in tetralogy of Fallot, despite it being a major driver of clinical decision-making. In particular, this study highlights the importance of the zones of apposition, pulmonary valve leaflet number, and pulmonary valve sinus number. Clinical evaluation now makes use not only of echocardiography, but increasingly CT and MRI. With care, all of the features we have identified should now be evident to those using these clinical techniques. Thus far, it has been metrics of the so-called pulmonary valvar “annulus”, in reality the virtual basal ring, that have been used to drive decision-making. Our study now provides evidence that other more specific features of the pulmonary root should be taken into consideration. We accept, nonetheless, that whilst our study offers novel data, it has its limitations. It was limited by the availability of specimens in the registries. There may also have been underlying bias in that all our patients had died. Thus, clinically milder phenotypes may not be well represented. This latter fact may well have influenced the proportions of hearts making up our clusters. The specific patterns we have identified, nonetheless, may have clinical significance. Only careful analysis in the clinical setting will show whether this is, indeed, the case. Additionally, it should be mentioned that the ability to discern clinical features is easier when examining a specimen versus examining a heart in situ using echocardiography, However, with the technological and the technical advances in echocardiography, the valve features discussed here should be able to be reproduced echocardiographically. Advanced imaging modalities such as computed tomography can also be utilised.