Pulmonary atresia with intact ventricular septum (PAIVS) is a rare disease with considerable morphologic diversity (1- 3). This has led to many different strategies for treatment. Outcome will depend, in great part, on the underlying cardiac morphology. Thus, reports of unusually good or disappointing results could reflect a selected population of patients with favorable or unfavorable pathology, rather than specific institutional or management factors. Owing to the rarity of this condition, more than nine-tenths of published series contain information on <30 patients. The range of morphology in a given study, therefore, may not be representative of the population of infants at large. The aim of our study was to provide a detailed description of the range of morphology in a population-based study (the United Kingdom and Ireland Collaborative Study of Pulmonary Atresia with Intact Ventricular Septum).

The Collaborative Study is a unique, contemporary, population-based study set up with the cooperation of all 18 pediatric cardiology centers in the United Kingdom and Ireland. From January 1, 1991, to December 31, 1995, all infants born with PAIVS and all fetal diagnoses of this condition were studied (4). Because of the high population density and small geographic area, it proved feasible for a single investigator (P. D.) to visit all 18 centers on repeated occasions. Local databases, admission and operative records and regional pathology records were directly inspected to ensure completeness of data collection. Cases were included if they had PAIVS without associated complex intracardiac abnormalities (cases with associated Ebstein’s malformation were included). Cases with tiny coexisting ventricular septal defects were included where the overall anatomy and physiology reflected that of PAIVS rather than pulmonary atresia with ventricular septal defect. Cases of critical pulmonary stenosis were excluded. Cases were also excluded if they were born outside the United Kingdom and Ireland. The cases diagnosed prenatally have been reviewed previously (4). Their range of morphology is not the subject of this report.

For each patient, the medical records, chest radiographs, electrocardiograms (ECGs), echocardiograms, hemodynamic findings, angiocardiograms and the operative and autopsy reports were directly reviewed. The echocardiogram at presentation was reviewed by a single investigator (P. D.) (Table le1). Tricuspid valve (TV) diameters and right ventricular (RV) inlet lengths, the latter taken from the mid-portion of the atrioventricular annulus to the apex of the RV, were measured by postprocessing from the cross-sectional echocardiographic four-chamber view at end-diastole (P. D., Z. S.) as previously described (5). Z-scores representing the number of standard deviations an observation deviates from the normal population mean for a given body surface area, were derived from echocardiographic data collected in 125 normal children (5) (Figure 1).

Each angiocardiogram was directly reviewed by three investigators (P. D., D. D., G. S.) (Table le1). The RV pressure obtained during initial cardiac catheterization was recorded. The degree of tricuspid regurgitation was estimated from review of the initial angiocardiogram and graded as absent, mild, moderate or severe. To allow statistical comparisons with other variables, patients were categorized into absent or mild, moderate or severe tricuspid regurgitation. Significant RV dilation was recorded when review of the angiocardiograms and echocardiograms demonstrated a hugely dilated and thin-walled ventricle. The angle that the patent arterial duct subtended on the postductal descending aorta was recorded from the echocardiogram as described by Santos et al. (6). The echocardiograms and angiocardiograms were reviewed to ascertain whether the atresia was due to muscular infundibular obliteration (muscular atresia), or complete fusion of the valve leaflets (membranous atresia), with a patent infundibulum existing in the latter setting up to the level of the valvar tissue.

Although appreciating that all three ventricular components are always present in this condition but with variable intracavity muscular overgrowth (7), the number of parts of the RV not obliterated by such muscular overgrowth (so-called tripartite, bipartite and unipartite 8- 9) was recorded. The presence of Ebstein’s malformation, and any additional morphologic abnormalities of the RV, was also recorded. The coronary arterial anatomy was closely studied from RV and aortic root injections, and the presence and position of RV-to-coronary fistulae and coronary arterial stenoses were recorded. When there was slight filling of nondilated coronary arteries from an RV angiogram, RV-coronary communications were termed “minor fistulae.” When there was prominent filling of one or more, usually dilated, coronary arteries associated with retrograde filling of the aorta, the communications were designated as “major fistulae.” Markedly dilated coronary arteries (>3 mm) were described as “ectatic,” and also considered as major fistulae. “RV dependent coronary circulation” was considered to be present when fistulous communications were associated with either absent aortocoronary connection, coronary arterial interruption or unequivocal stenosis of one or more of the major epicardial coronary arteries. Markedly ectatic coronary arteries were also considered in this category, as RV decompression could result in major coronary steal through the ectatic vessel to the RV (10).

Autopsy reports for all the deceased patients were available for review (Table le1) and, in addition, the cardiac specimen was reexamined for the purposes of this study in 50% of cases by one of the investigators (R. A.), or by a regional pediatric pathologist who recorded findings on a standardized questionnaire. Overall morphology at, or prior to, initial intervention was ascertained by combination of the information from echocardiograms, angiocardiograms, operative notes, postmortem records and postmortem specimens (Table le1). Associated cardiac and noncardiac anomalies were also recorded. Throughout the results, the frequencies of specific anatomical findings are expressed as a proportion of the total number of cases for which adequate information was available (Table le1).

The statistical significance of two morphologic features coexisting in the same heart was examined using Pearson’s chi-square test with Yates’ correction when expected values were small, the Student t test and linear regression depending on whether the data were categoric or continuous. This was expressed as a p value (Table le2). P values <0.05 were considered statistically significant. For subgroups with various significant morphologic characteristics, a 95% confidence interval (CI) was calculated; Z-scores for the TV and RV inlet were expressed as cumulative frequency distributions. All data was stored in coded fashion on a relational database (Filemaker Pro 3.0, Claris, Santa Clara, California), and all analyses were performed using a commercial statistical package (Statview 4.0, Abacus Concepts, Berkeley, California).

From 1991 to 1995, there were 183 infants born alive with PAIVS in the United Kingdom and Ireland, giving an overall incidence of 4.5 cases per 100,000 live births. Distribution of births has been described previously (4). There were 86 fetal diagnoses during this period, leading to 53 terminations of pregnancy (61%). Of the 183 live births, 70 have died postnatally.

Median TV Z-score at presentation was −5.2 (interquartile range −8.4 to −2.6) (Figure 2) as derived using our own normal echocardiographic data (5) (Figure 1) or −1.6 (−2.9 to −0.4) using Rowlatt’s “normal” data based on postmortem measurement of formalin-fixed samples (11). The relationship between TV Z-score and other morphologic variables is shown in (Table le2). Ebstein’s malformation coexisted in 18/183 cases (9.8%); in 10 there was mild, and in 8 moderate or severe, displacement of the septal leaflet of the TV.

Significant RV dilation was present in 8/183 cases, and this was associated with moderate to severe tricuspid regurgitation in all cases (Figure 3). Of the eight cases, Ebstein’s malformation was present in three, marked TV dysplasia in two, and a seemingly normal TV, but with a dilated orifice, in the remaining three. The RV was exceptionally thin-walled in four of the eight cases. Hypoplastic pulmonary arteries were present in all but one of the eight. None had ventriculocoronary connections.

Median RV inlet length Z-score at presentation was −5.1 (interquartile range −7.5 to −2.8) (Figure 4) as derived using our own normal echocardiographic data (5) (Figure 1) or −5.3 (−7.6 to −3.0) using the normal echocardiographic data of Hanseus et al. (12). The relationship between RV inlet Z-score and other morphologic variables is shown in (Table le2).

Of the 143 where it was definitely possible to ascertain the RV morphology, muscular obliteration of both apical and infundibular portions of the RV resulted in so-called “unipartite” RV in 7.7% of cases (95% CI, 3.3% to 12.1%) (Figure 3). Overgrowth of the apical trabecular portion resulted in “bipartite” ventricles in 33.6% of cases (95% CI, 18.6% to 48.6%) (Figure 3). There were no examples of infundibular obliteration without apical obliteration. The remaining 58.7% (95% CI 50, 1% to 67.4%) had “tripartite” morphology (Figure 3). The atresia was membranous in 74.7% of cases (95% CI, 68.3% to 81.2%) and muscular in 25.3% (95% CI, 18.2% to 32.4%) (n = 174).

There was adequate visualization of the coronary arteries in 131 patients by angiography or at autopsy (Figure 3). The coronary arteries were normal in 71 patients (54.2%) (95% CI, 45.7% to 62.7%), including 16% with prominent but blind-ending fissures in the ventricular wall that filled in ventricular systole without communicating with the coronary arteries. Minor fistulae were present in 28 patients, and major fistulae in 33 patients, giving a total number of 61 (45.8%) (95% CI, 37.3% to 54.3%) with fistulous communications out of 131 patients. The coronary arterial regions where fistulous connections occurred, and their frequency of occurrence, are documented in the Appendix.

Ten cases, 7.6% (95% CI, 3.1% to 12.2%), were considered to have an “RV-dependent” coronary circulation. Stenosis, interruption or severe ectasia of the right coronary alone was documented in five cases (3.8%), of the left coronary alone in two (1.5%), and of both coronary arteries in another two cases (1.5%). In one case with major fistulous connections to the aorta, there was insufficient information to localize the site of the RV dependence. Fistulous connections were most commonly to the distal right and proximal left anterior descending (LAD) coronary arteries, and rarely to the circumflex artery, when they were extremely distal (Appendix). In one case, there was a connection via the coronary arteries between the RV and pulmonary arteries. Stenoses occurred predominantly in the distal right and distal LAD arteries, usually close to the site of entry of a fistula. Ectasia was found equally in all parts of the right coronary artery and/or LAD, and was often associated with significant luminal irregularity.

One case studied at autopsy had both an atretic orifice of the right coronary artery and absence of the orifice of the left coronary (Figure 5). The right coronary arterial orifice was represented by only a dimple, and there was no evidence of another orifice. The main stem of the left coronary artery was also absent, and the coronary supply was provided by large fistulae from the RV to the distal right coronary artery, the distal LAD, and the distal left circumflex artery (Figure 5). Other variants studied by angiography and at autopsy included one case of the right coronary artery originating from the left coronary artery, and another of both coronary arteries originating from a left-sided sinus.

Fistulae tended to occur in the smaller ventricles as judged by TV Z-score (p < 0.0001) and RV inlet Z-score (p = 0.0011). Relations with other variables are shown in (Table le2). Of those patients with fistulae, the presence of stenoses, interruptions and ectasia was related to higher RV pressure recorded at catheter prior to initial procedure but no other variables (Table le2). Of 10 cases with RV dependence, 8 cases had membranous atresia and 2 had muscular atresia. In terms of ventricular structure, six cases had so-called tripartite ventricles, three bipartite, and one case was unknown. The largest TV and RV inlet Z-score documented in such a ventricle was −1.7 and +3.5, respectively.

The presence of RV-to-coronary fistulae, as outlined above, was documented before intervention. Subsequent to surgery, there were examples of change in the presence or degree of fistulae. There were three cases of regression of fistulae to a normal coronary arterial arrangement following RV decompression, and two cases of RV-to-coronary fistulae developing stenoses. Neither of these patients had undergone RV decompression.

The angle that the arterial duct subtended with the postductal region of the aorta was obtuse (normal) in 56/112 cases (50%) (95% CI, 40.7% to 59.3%) where it could be ascertained, at right angles in 11 cases (9.8%) (95% CI, 3.95% to 14.6%) and acute in 45 cases (40.2%) (95% CI, 31.1% to 49.8%) (Figure 6). Acute-angled ducts also tended to arise more proximally on the aortic arch than normal. The pulmonary arteries were always confluent. Aortopulmonary collateral arteries were present in three cases, one with multiple collaterals as the major source of pulmonary blood flow with miniaturized confluent pulmonary arteries and tiny left-sided duct making little contribution. In the other two cases, small collateral arteries were present, but the major source of pulmonary blood flow was via the arterial duct into the pulmonary arteries. Sixteen infants had significantly hypoplastic (<3 mm) pulmonary arteries as assessed by echocardiography and/or angiocardiography (Table le3). The aortic arch was right-sided in three cases with a left-sided duct in one case (unknown in the other two).

These are documented in (Table le3). Fifteen abnormalities of the left ventricle (LV), the subaortic outlet and the mitral valve were found in 11 patients. Although it is acknowledged that many cases of PAIVS have septal bowing into the LV outflow tract, and that this may represent a spectrum, there were four cases where this was extremely pronounced.

Review of the surgical literature shows that morphologic information is often scant. This hinders comparison between studies, making it difficult for the reader to determine whether the population described is representative of the population of PAIVS as a whole. Our study describes the range of morphology in an unselected population. This information can now serve as a standard for other studies to ensure that groups are representative of the population of patients with pulmonary atresia with intact septum as a whole. In this manner, it may be possible to decide whether studies with particularly high survival rates represent real advances in management, or whether the population under review is biased in favor of patients with less severe pathology.

Our population-based morphologic data is broadly comparable with the larger nonpopulation-based studies (13- 14). Using Rowlatt’s postmortem data to generate Z-scores to aid comparability (11), the median TV Z-score for this study was −1.6. It was −2.2 in the Congenital Heart Surgeons Study (CHSS) (13) and −4.8 in the study reported by Bull et al. (14). We do not, however, advocate use of normal postmortem data to generate Z-scores for echocardiographic measurements of individual patients. Shrinkage occurs to fixed autopsy material, and normal dimensions derived at autopsy will be smaller than those derived from echocardiography. Use of autopsy—rather than echocardiographically derived normal dimensions to generate Z-scores—will, therefore, lead to relatively large (less negative) median Z-scores. Because cross-sectional echocardiography is now the principal tool used for evaluation of infants with PAIVS before initial intervention, we recommend that Z-scores be calculated using normal data based on the same measurement techniques (5).

In our study, 45.8% of patients had RV-coronary arterial fistulae, virtually identical (p = 0.90) to the proportion of 45% found in the CHSS (13). In the Toronto series, the proportion was 56% (p = 0.13 compared to our proportion) (15), but only 32% (p = 0.05) in Boston (16). Our incidence of RV-dependent coronary circulation, at 7.6%, was similar (p = 0.83) to the 9% found by the CHSS. In the Boston study (16), one or more stenoses occurred in 16/82 (19.5%) patients studied (p = 0.01). The distribution of coronary arterial fistulae was broadly similar to the findings of the Boston group (16). Unlike their study, we found stenoses predominantly in the distal right coronary artery and distal LAD. The Boston group found stenoses most commonly in the mid-LAD (16).

Fistulae tended to occur in patients with the smaller RVs as judged by TV size and RV inlet length, but did occur across the range of ventricular size, and they were present in a quarter of patients with so-called tripartite ventricles with membranous atresia (Figure 2). Hanley et al. (13) found fistulae correlated with smaller TV Z-score, less TV incompetence, and higher RV systolic pressure. Rychik and colleagues (17) documented that patients with more severe coronary abnormalities had smaller TV Z-scores. We, in addition, showed correlation with smaller RV inlet dimension Z-score, acute-angled duct, muscular atresia, and so-called unipartite and bipartite RV. An RV-dependent coronary circulation (stenoses, interruptions or severe ectasia) correlated only with RV pressure at presentation and not with any marker of ventricular size. A recent study from Boston also failed to demonstrate a statistically significant relationship between RV-dependent coronary circulation and the degree of RV hypoplasia (18). The CHSS did demonstrate a correlation with smaller TV Z-score (13).

Absence of the coronary arterial orifices is a rare but described occurrence in PAIVS. Lenox and Briner (19) first reported a case in a two-month-old who lacked proximal aortocoronary connections with a coronary circulation entirely RV dependent. We found only a single case in our series. This infant died within 24 h of birth despite infusion of prostaglandin.

An epicardial dimple was found in one patient at surgery overlying the trabecular portion of RV and close to the LAD. It has been suggested that such dimples may represent subepicardial coronary arteries (20), and may be the external stigmata of coronary artery fistulae indicating the site of such connections (3). The patient in our study was not known to have any fistulae.

The angle that the arterial duct subtends with the postductal descending aorta has been used by some groups as an indicator of whether the pulmonary outflow became atretic earlier or later in gestation (6,21- 22). In our study, the ductal angle was acute in 40.2% of cases compared to 45% in the PAIVS patients in the Kutsche et al. study (21) (Figure 6). This tended to occur in those with smaller ventricles with muscular atresia, and ventriculocoronary arterial connections. It may be that this subgroup represents an earlier occurring lesion compared to those with well-formed ventricles, which may have progressed from pulmonary stenosis to atresia in the last trimester of pregnancy (4,23). The duct provided the main source of pulmonary blood flow in almost all our patients. Two patients had additional small aortopulmonary collateral arteries. Only in one case did multiple aortopulmonary collaterals provide practically all the pulmonary blood flow. This is uncommon, but it has been documented previously (24).

Abnormalities of the structure of both the LV and mitral valve have been described in PAIVS (1,25- 26). In a postmortem study, Akiba and Becker (26) found that in half of the cases studied, the tendinous cords of the mitral valve were dysplastic and shortened. All LVs were hypertrophied and showed an increase in interfibre collagen, suggesting chronic ischemia (26). In our series, a lower incidence of mitral valvar abnormalities (1.6%) was detected, possibly because most did not come to postmortem examination. Concentric LV hypertrophy was present in only a single patient. Convex bulging of the LV septal surface has been described by Zuberbuhler and Anderson (1), and subsequently by Freedom (25) and Akiba and Becker (26) in patients with small and hypertensive RVs. The subaortic bulge has been reported as promoting severe LV outflow obstruction after the Fontan operation in PAIVS (27), but before this procedure it is highly unusual (17). Subaortic obstruction due to accessory TV tissue prolapsing into the LV outflow has been documented in children with raised RV pressures and septal defects such as complete transposition with ventricular septal defect (VSD), and pulmonary stenosis and VSD (28). To our knowledge subaortic stenosis has not been documented in PAIVS as a consequence of accessory TV tissue walling off a VSD and prolapsing into the LV outflow due to suprasystemic RV pressures.

Noncompactile spongiform myocardium is characterized by a trabecular network with wide anastomosing sinusoidal blood channels giving a honey-combed, sponge-like appearance on angiography. It can occur in isolation (29- 30), or in the setting of PAIVS (31). When it occurs in the presence of PAIVS, it can be present in either ventricle including in the ventricular septum, as in the case we report (Table le3).

Although the inclusion of tiny ventricular septal defects in our series of pulmonary atresia with intact ventricular septum (Table le3) may seem contradictory, the overall anatomy and physiology reflects that of PAIVS rather than pulmonary atresia with ventricular septal defect. Small ventricular septal defects have been described before in PAIVS, both in prenatal and postnatal life (4,32).

Abnormalities of the aortic valve have been reported in PAIVS, including critical aortic stenosis (33), but not unicuspid and unicommissural aortic valves. This rare subtype of aortic stenosis, most commonly found in neonates, is rarely an isolated lesion, and it is frequently associated with disorders of the left heart (34). Occurrence of bicuspid aortic valve in PAIVS seems to mirror that in the general population. Its incidence at large has been estimated at 0.4% to 2.0% (35- 37). Our study documented the incidence in PAIVS at 1.6%.

The relationship between undue prominence of the venous valves, in particular the Eustachian valve, and pulmonary atresia was observed by Kauffman and Anderson (38), who speculated that they may be causal in right-heart hypoplasia. They were only found in 3 of the 183 cases in our series. Venous return is usually normal. The incidence of a persistent left superior caval vein was 2.7%, and the vessel always drained to the coronary sinus. This compares to about 4% in patients with congenital heart disease in general, and 0.3% in the general population (39). Unilateral pulmonary venous stenosis in the setting of PAIVS also seems to be exceedingly rare. It has been observed by the Toronto group as a rare association of PAIVS (40).

The population-based range of morphology portrayed in this study was ascertained at birth, or soon after. Fetal diagnosis is now widespread in the United Kingdom, and the incidence of the disease at birth has been reduced by 23% on average owing to selective termination of pregnancy (4). Various morphologic subgroups of pulmonary atresia might be detected more easily in fetal life, and selective termination could affect the spectrum of morphologic features seen at birth. We have shown previously, however, that there were no morphologic differences at birth between patients who had been diagnosed antenatally compared to those who had not (4). This would imply that any such effect, if present, is probably small.

The ascertainment of RV-coronary arterial fistulae depends, for the most part, on filling of the coronary arteries following a RV angiocardiographic injection. Additional information may also be obtained by high-quality aortography. Technical differences—for example, hand versus power injections—may create marked variations in quality of delineation of coronary abnormalities. In the current study, we optimized our ability to define anatomy fully by review of all pertinent information, including echocardiography and angiocardiography, and surgical and autopsy findings.

This was an observational study. As such, no attempt was made to influence management. This resulted in variable investigation of each patient. Thus, complete data was not available on all patients despite the investigators’ best efforts. For example, not all patients had a preoperative angiocardiogram, nor was an autopsy performed on all patients who died.

The United Kingdom and Ireland Collaborative Study of Pulmonary Atresia with Intact Ventricular Septum is one of the first population-based studies to provide comprehensive data on a large cohort of infants with a rare congenital cardiac malformation. Unlike disease registries, we have not relied on local investigators to forward data to the coordinating center. A single investigator visiting each center on multiple occasions ensured completeness of collection of data and allowed key diagnostic studies, such as echocardiography and angiocardiography, to be directly reviewed and recorded for subsequent review.

We have described the range of morphology in a population-based study of pulmonary atresia with intact ventricular septum. This may serve as a reference source for clinical studies to judge whether their morphology is representative of the population at large.

The authors wish to thank all the pediatric cardiologists and pediatric cardiac surgeons at the participating institutions, as well as regional pathologists: Alder Hey Children’s, Liverpool; Birmingham Children’s, Birmingham; Bristol Children’s, Bristol; Freeman, Newcastle; Glenfield, Leicester; Guys and St. Thomas’, London; Harefield, Middlesex; Hospital for Sick Children, Great Ormond Street, London; John Radcliffe, Oxford; Killingbeck, Leeds; Our Lady Hospital for Sick Children, Dublin; Royal Belfast Hospital for Sick Children, Belfast; Royal Brompton and National Heart, London; Royal Hospital for Sick Children, Edinburgh; Royal Hospital for Sick Children, Yorkhill, Glasgow; Royal Manchester Children’s, Manchester; University Hospital of Wales, Cardiff; and Wessex Cardiothoracic Centre, Southampton. We also thank Dr. Yen Ho for the preparation of the specimen in (Figure 5), and Karen McCarthy for preparation of the figures and the manuscript.