Discrete subaortic stenosis represents a form of left ventricular outflow tract obstruction characterised by a membranous ridge proximal to the aortic valve.Reference Stassano, Di Tommaso and Contaldo1 Although rare cases of neonatal discrete subaortic stenosis have been reported,Reference El Habbal2, Reference Kleinert, Ott and Geva3 this lesion more commonly presents as an acquired defect in early infancy and accounts for 8–30% of left ventricular outflow tract obstruction in children.Reference Kirklin and Barratt-Boyes4, Reference Rayburn, Netherland and Heath5 Risk of post-operative aortic insufficiency, the potential for recurrent stenosis, and surgical management of discrete subaortic stenosis have traditionally been fairly conservative due to the unpredictable rate of progression, postponing resection until a certain left ventricular outflow tract gradient has been reached. However, earlier surgical intervention is now advocated by several centresReference Rayburn, Netherland and Heath5–Reference Parry, Kovalchin and Suda8 in an effort to avoid progressive left ventricular hypertrophy, aortic insufficiency, and endocarditis, which can be associated with discrete subaortic stenosis.Reference Feigl, Feigl, Lucas and Edwards9–Reference Wright, Keane, Nadas, Bernhard and Castaneda12 This earlier intervention can only be advocated now that the risk of post-operative morbidity and mortality at many centres is minimal.
Although increasing left ventricular outflow tract gradient is a well-accepted indication for resection of a discrete subaortic shelf, this indication may overlook the shelf’s progressive nature to grow and encroach on the leaflets of the aortic valve, potentially impairing future valve function. Transthoracic echocardiography is used to diagnose and follow the progression of left ventricular outflow tract obstruction; however, its ability to demonstrate the extent of discrete subaortic shelf encroachment onto the aortic valve is unclear. If transthoracic echocardiography is unable to determine involvement of the aortic valve except perhaps for the late finding of aortic insufficiency or stenosis, resection on diagnosis may be warranted, given the low rate of post-operative morbidity associated with isolated discrete subaortic shelf resection. The purpose of our study was to determine the sensitivity and specificity of transthoracic echocardiography for identifying involvement of the aortic valve by a discrete subaortic shelf.
Methods
A retrospective review of the cardiac surgery database at Texas Children’s Hospital from October 1995 to May 2006 was conducted for patients undergoing primary repair for discrete subaortic stenosis. We identified 48 patients, of whom 41 (85%) had isolated discrete subaortic shelf resection and 7 (15%) had concomitant closure of an atrial septal defect. Patients undergoing any other concomitant procedures such as ventricular septal defect closure were excluded.
Analysis of aortic valve encroachment
Pre-operative transthoracic echocardiography was performed in 47 patients (98%) and 6 patients (13%) underwent pre-operative cardiac catheterisation (Table 1). The diagnosis of aortic valve encroachment by the discrete subaortic shelf was determined from the official pre-operative echocardiogram reading and from a retrospective blinded review of the pre-operative echocardiograms by an independent echocardiographer. Since the blinded second reading was looking specifically for involvement of the aortic valve by the shelf, this would theoretically maximise the sensitivity and specificity of transthoracic echocardiography for diagnosing any involvement of the aortic valve. These data were then compared with the surgeon’s intra-operative findings.
Table 1 Indications for cardiac catheterisation (n = 6).

Statistical analysis
Non-parametric, binomial, and descriptive statistics were computed when appropriate. Data were described as ratios, medians with ranges, and means with standard deviations. The Fisher exact test and the χ2 test were used to analyse binary variables. Kaplan–Meier freedom from re-operation analysis plotted the re-resection curve from the follow-up data. For this analysis, children were censored at the time of re-resection or were withdrawn free from re-operation at the point of last contact. Freedom from re-operation time was defined as age in months with date of surgery taken to be time zero (Fig 1).

Figure 1 Kaplan–Meier freedom from re-operation after discrete subaortic shelf resection.
Sensitivity and specificity ratios were calculated to assess the ability of the transthoracic echocardiogram to diagnose involvement of the aortic valve by the discrete subaortic shelf. Multivariable analysis was performed using the Cox proportional hazard model. Four variables entered the models independently and not stepwise. They were specified and tested for significance as risk factors for recurrence of obstruction requiring re-resection of the discrete subaortic shelf. Continuous variables tested were pre-operative peak left ventricular outflow tract gradient and age, in years, at primary resection. Discontinuous variables were pre-operative aortic insufficiency and septal myectomy. All analyses were conducted with SPSS 13.0 (SPSS, Inc., Chicago, IL, USA).
Conduct of the operation
All operations were performed utilising cardiopulmonary bypass, with mean bypass and aortic cross-clamp times of 90 plus or minus 28 minutes and 60 plus or minus 21 minutes, respectively. Access to the discrete subaortic shelf was achieved through a transverse aortotomy. Sharp, circumferential resection of the entire obstructing shelf was performed. This technique includes stripping the shelf off the mitral valve and aortic valve leaflets with the goal of restoring the leaflets to their normal thickness and mobility. Myectomy of the hypertrophied interventricular septum was performed in 37 patients (69%).
Results
Patient demographics
The mean age was 7.2 plus or minus 6.0 years and the mean body surface area was 1.9 plus or minus 0.4 square metre. A total of 29 patients (60%) were males. Altogether 10 patients (21%) had a bicuspid aortic valve and four (8%) had additional non-cardiac congenital anomalies, with one patient each having Alagille’s Syndrome, a variant of Shone’s Syndrome, Teebi’s hypertelorism syndrome, and unspecified multiple congenital anomalies. A total of three patients had previous surgery to repair coarctation of the aorta, and one patient underwent previous ligation of a patent ductus arteriosus; all were performed through left thoracotomy.
Analysis of aortic valve encroachment
Intra-operative evaluation revealed involvement of the aortic valve by a discrete subaortic shelf in 65% (31/48) of patients. The pre-operative echocardiogram reading identified involvement of the aortic valve in 11 of 31 patients (35%) in whom this was present, while the blinded echocardiogram readings identified involvement of the aortic valve in 10 of 31 (32%) of these patients. A total of 65% of the cases with the involvement of the aortic valve went undiagnosed by pre-operative transthoracic echocardiography. There was a significant difference between involvement of the aortic valve predicted by both the official pre-operative echocardiogram reading and the blinded review of the pre-operative echocardiogram when compared with the involvement of the aortic valve documented at surgery (p < 0.01, p < 0.01). The sensitivity and specificity of pre-operative transthoracic echocardiography to diagnose involvement of the aortic valve are 35% and 76% (Table 2).
Table 2 Sensitivity and specificity of pre-operative transthoracic echocardiography for predicting involvement of the aortic valve by a discrete subaortic shelf.

Echo, echocardiogram
Pre-op: official pre-operative echocardiogram; blinded: independent echocardiographer’s blinded review of official pre-operative echocardiogram
Involvement of the aortic valve: pre-op echo 23% (11/48) versus intra-operative findings 65% (31/48); p < 0.01
Sensitivity/specificity of pre-op echo to diagnose involvement of the aortic valve = 0.35/0.76
Involvement of the aortic valve: blinded echo 21% (10/48) versus intra-operative findings 65% (31/47); p < 0.01
Sensitivity/specificity of Blinded Echo to diagnose involvement of the aortic valve = 0.32/0.94
Pre-operative transthoracic echocardiography demonstrated a mean peak left ventricular outflow tract velocity of 4.1 plus or minus 0.8 milliseconds with a mean peak left ventricular outflow tract gradient of 69.1 plus or minus 26.2 millimetres of mercury. Aortic insufficiency was trivial or absent in 16 patients (34%), mild in 30 patients (64%), and moderate in 1 patient (2%). Altogether 22 (71%) of the 31 patients with pre-operative aortic insufficiency had the involvement of the aortic valve by the discrete subaortic shelf at surgery.
Follow-up
Mean follow-up was 3.4 plus or minus 2.9 years and was 100% complete. All six patients (13%) required re-resection of the discrete subaortic shelf, all of whom had involvement of the aortic valve at their first resection. All six re-operations were successful with no morbidity or mortality. Of the six patients who had re-operations, the official and blinded pre-operative echocardiogram readings only predicted the involvement of the aortic valve in one patient.
At follow-up, one patient who had undiagnosed involvement of the aortic valve before primary resection later underwent a Ross procedure for aortic insufficiency and stenosis. Three additional patients (6% of the entire cohort) went on to develop late moderate aortic insufficiency, 6 months and above after discrete subaortic shelf resection, all of whom left the operating room with mild aortic insufficiency. Two of these patients had involvement of the aortic valve at their primary surgery, neither of which were identified by pre-operative transthoracic echocardiography.
Freedom from re-operation or death and risk factors for recurrent stenosis
Kaplan–Meier freedom from re-resection was 100% at 1 year, and 88% at 6 years (Fig 2). Of the four variables analysed as risk factors for recurrent stenosis, younger age was the only statistically significant risk factor for re-resection (p < 0.05), consistent with previously published findingsReference Brauner, Laks, Drinkwater, Shvarts, Eghbali and Galindo6 (Table 3). All re-resections were performed in patients who underwent primary resection before the age of 7 years, with five patients being below the age of 4 years. Contrary to the earlier series,Reference Serraf, Zoghby and Lacour-Gayet13 none of the three patients who underwent previous coarctation repair have required re-resection.

Figure 2 Pre-operative transthoracic echocardiogram demonstrating a discrete subaortic shelf.
Table 3 Risk factors for re-stenosis by a discrete subaortic shelf.

LVOT, left ventricular outflow tract
Early and late survival was 100%. There were no strokes or re-operations for bleeding. None of the 48 patients developed wound infection or required pacemaker implantation. There were no cases of post-operative renal insufficiency requiring dialysis. The mean hospital length of stay was 4.4 plus or minus 1.6 days.
Discussion
First described by CheeversReference Cheevers14 in 1842, discrete subaortic stenosis appears to be an acquired lesion with either fibrous or fibromuscular subtypes.Reference Kelly, Wulfsberg and Rowe15 Associated sequelae include infective endocarditis, aortic insufficiency, left ventricular hypertrophy, and recurrence after surgical resection.Reference Feigl, Feigl, Lucas and Edwards9–Reference Wright, Keane, Nadas, Bernhard and Castaneda12 Although an increasing left ventricular outflow tract gradient greater than 25–40 millimetres of mercury is commonly accepted as an indication for surgery, several controversies exist regarding the optimal surgical management of these patients, including additional indications for shelf resection, the timing of resection, and the indications for septal myectomy. Additional indications for resection at many centres have come to include the presence of co-existing cardiac lesions and evidence of new or progressive aortic insufficiency. The progressive nature of discrete subaortic stenosis is unpredictable, thus the timing of surgical intervention is controversial. Delaying resection until the patient demonstrates progression of left ventricular outflow tract obstruction avoids the inherent risks of surgery in patients whose left ventricular outflow tract obstruction is mild and non-progressive. On the other hand, some centres, including ours, recommend earlier surgical referral in attempting to avoid the development of left ventricular hypertrophy and, most importantly, aortic insufficiency. Discrete subaortic shelf in the vast majority of cases is a progressive lesion and waiting until there is significant left ventricular outflow tract obstruction or the late finding of aortic insufficiency, misses the opportunity to avoid surgery on the aortic valve. Scraping a fibrotic shelf off the aortic valve is technically challenging and puts the aortic valve leaflets at risk of perforation. In addition, it is difficult to imagine that this damage to the endothelium of the aortic leaflets does not change the natural history of the valve’s function. Furthermore, since many centres can perform this procedure with almost no morbidity or mortality as demonstrated in the current series, there are few disadvantages of proceeding to surgery on diagnosis. However, this current series as well as others have demonstrated that earlier age at resection is associated with a higher re-operative rate. Concomitant myectomy is advocated by several groupsReference Rayburn, Netherland and Heath5, Reference Wright, Keane, Nadas, Bernhard and Castaneda12, Reference Hardesty, Griffith and Mathews16, Reference McKay and Ross17 including ours to achieve maximum relief of left ventricular outflow tract obstruction and perhaps reduce re-growth of the discrete subaortic shelf by changing the angle between the left ventricular outflow tract and the body of the left ventricle. However, others have reported little additional benefit with septal myectomy.Reference Ashraf, Cotroneo and Dhar18
This series demonstrates that transthoracic echocardiography has neither the sensitivity nor specificity to reliably demonstrate aortic valve encroachment by a discrete subaortic shelf. It is possible that transesophageal echocardiography may improve the sensitivity and specificity of the involvement of the aortic valve; however, this modality is invasive and thus not without the associated risks inherent to any invasive procedure. We do not recommend transesophageal echocardiography as a routine modality for following discrete subaortic shelf progression due to its invasive nature. An example of a discrete subaortic shelf that on the pre-operative transthoracic echocardiogram did not appear to involve the valve but did at operation is shown in Figure 2. Furthermore, using patients’ left ventricular outflow tract gradient as the primary indication for surgical intervention may allow aortic valve encroachment to go unnoticed, increasing the risk of functional impairment to the aortic valve in the future. All six patients in our series that required re-operation had involvement of the aortic valve. Given that younger age at the time of primary resection was found to be a significant risk factor for re-resection, with five of the six patients requiring re-operation being below the age of 4 years, we recommend considering elective resection of all discrete subaortic shelves at the time of initial diagnosis for patients above the age of 4 years.
Once the aortic valve has been affected by a discrete subaortic shelf, resection of the lesion with debriding and stripping of the shelf off the valve leaflets will often improve moderate or severe aortic insufficiency and/or aortic stenosis, but this technique does not make the valve normal.Reference Stassano, Di Tommaso and Contaldo1 This extensive manipulation of the aortic valve leaflets to remove an adherent shelf likely compromises the aortic valve leaflets’ endothelium and deeper layers, which most likely changes the natural history of these valves, perhaps resulting in an increased risk of late aortic insufficiency and aortic valve replacement. This damage to the valve may also explain the common persistence of mild aortic insufficiency post-operatively (27%). In this series, three patients developed late post-operative moderate aortic insufficiency. Two of these patients had a discrete subaortic shelf adherent to the aortic valve at the time of the operation, which was not identified by pre-operative transthoracic echocardiography. It is the opinion of the authors that dissecting an adherent shelf off the aortic valve alters the natural history of the valve and that the risk of such valvular involvement by a progressively growing shelf outweighs the risks associated with primary resection of an isolated discrete subaortic shelf. Since all six patients in our series, who required re-resections, had the involvement of the aortic valve at the time of their primary resection, the risk of future re-operation associated with the involvement of the aortic valve must also be accounted for. Longer follow-up is required to clearly delineate the presumption that discrete subaortic shelf involvement of the aortic leaflets with subsequent surgical removal of that fibrosis does change the natural history of the aortic valve.
Conclusions
Transthoracic echocardiography is unreliable in demonstrating involvement of the aortic valve by a discrete subaortic shelf due to its relatively poor sensitivity and specificity. Following left ventricular outflow tract gradients by echocardiogram may allow the involvement of the aortic valve by a discrete subaortic shelf to go unnoticed. Elective resection of all discrete subaortic shelves at the time of initial diagnosis for the patients above the age of four should be considered because of the risk of future aortic valve dysfunction and the negligible morbidity and mortality associated with primary resection.