Resection of subaortic stenosis; can a more aggressive approach be justified?

Abstract

Objectives: Discrete subaortic stenosis causes left ventricular outflow tract (LVOT) obstruction and often produces aortic regurgitation (AR) which alone may precipitate surgical intervention. Conventional resection relieves the obstruction, but the recurrence rate is high, and the AR is little changed as the thick fibrous membrane which extends onto the valve leaflets remains. We studied whether an aggressive surgical approach could reduce both the severity of AR and rate of recurrence of obstruction associated with discrete subaortic stenosis, and whether this aggressive approach could be justified. Methods: Between June 1992 and April 1996, 37 patients aged 0.5–35 years (median 7.5) underwent resection of a discrete subaortic membrane. Ten underwent re-operation for recurrent obstruction and eight followed previous ventricular septal defect closure. LVOT gradient was measured using the modified Bernoulli equation and AR was graded on a scale of 0–4 (0=none, 4=severe). Postoperative assessment was performed early (≪7 days) and at mid-term (27.0 months; range 2–59 months). Results: There was significant improvement in AR from mild/moderate to none/trivial (P=0.019) immediately postoperatively and LVOT gradient from 66.9±30.4 to 15.1±12.2 mmHg (P≪0.0001). By stepwise logistic regression preoperative gradient correlated significantly with postoperative mild/moderate AR (P=0.015) and LVOT gradient (P=0.0036). Preoperative mild/moderate AR also correlated with postoperative mild/moderate AR (P=0.034). Five patients developed complete heart block, four undergoing reoperation for recurrent obstruction, and one preoperatively had right bundle branch block from previous ventricular septal defect repair. At mid-term follow-up there was no increase in AR or LVOT gradient (14.8±12.8 mmHg). Early post-operative AR was the strongest predictor of late mild/moderate AR (P=0.02). Early post-operative gradient was a weaker predictor (P=0.04). Pre-operative and early post-operative gradient were significant predictors of late gradient (P=0.0038; ≪0.0001, respectively). No patient required reoperation for recurrent obstruction; one underwent late aortic valve replacement for severe AR. Conclusions: An aggressive surgical approach to discrete subaortic stenosis produces excellent relief of obstruction and frees the valve leaflets, significantly reducing associated AR at early and mid-term follow-up with low morbidity for primary operation. Long-term follow-up is required to confirm whether this early benefit is maintained.

Subaortic stenosis, Aortic regurgitation, Recurrent obstruction

Introduction

Surgery for discrete subaortic stenosis is associated with a high incidence of late problems which often require further surgical intervention. The two most significant late problems are aortic regurgitation (AR) which may occur up to 12 years after the initial surgery and recurrence of the left ventricular outflow tract (LVOT) obstruction.

Attempts to reduce the incidence of late complications are hampered by our lack of understanding of the cause of the sub-aortic obstruction. Indeed, the single entity considered may be a heterogeneous collection of pathological abnormalities. The obstructive element itself is due to two factors. First, there is the actual sub-aortic fibromuscular ridge. Second, the fibrous tissue which coats and tethers the leaflets of the aortic valve limits its opening and thereby exacerbates the LVOT obstruction (Fig. 1 ). From the fact that the fibrous ridges are not apparent at birth, they appear to be acquired and related to a disturbance in blood flow, thus supporting the contention that hemodynamics play a part in their development [1],[2]. An opposing theory proposed by Feigl et al. [3] particularly relevant to the valvulopathy and the obstruction due to this, is that a continuous fibroelastic membrane develops in the LVOT and progresses upwards to involve the leaflets of the aortic valve as well.

Fig. 1

Echocardiogram of a typical patient with 'discrete' subaortic stenosis. Involvement of the leaflets of the aortic valve by the obstructing subaortic membrane can clearly be seen with encroachment of subaortic fibrous tissue onto the aortic valve leaflets. The mass of fibrous tissue obviously contributes to obstruction to the outflow tract from the left ventricle.

The AR appears to be due to thick fibrous tissue on the left ventricular surface of the leaflets of the aortic valve. Whether this is due to repetitive trauma from a jet of blood through the region of the stenosis striking the leaflet [1],[2], or due to development of the fibroelastic membrane postulated by Feigl [3], it is perceived that the fibrous tissue develops, then by scarring and retraction deforms the leaflets of the valve resulting in insufficiency.

Conventional surgical approaches have mainly focused on the obstructive element of the lesion and treatment of the AR has assumed secondary importance. However, it is reasonable to postulate that if the cause of the altered hemodynamics in the LVOT could be completely removed, and that if the fibrous membrane could be excised in its entirety the severity of post-operative AR would be less and the frequency of late complications would be reduced. We therefore pursued a very aggressive surgical approach to discrete subaortic stenosis in an attempt to address both of these issues and studied whether this aggressive approach could be justified in terms of improvement in AR in the short-term, and reduced incidence of long-term complications overall.

Patients and methods;

Patients

Between June 1992 and April 1996, 40 patients of median age 7.5 years (ranged from 6 months to 35 years) underwent resection of discrete subaortic stenosis as part of the surgical procedure. Subaortic stenosis was defined as a discrete fibrous membrane or fibromuscular ridge in the LVOT as seen echocardiographically. Patients with significant valvar aortic stenosis, tunnel LVOT obstruction, or idiopathic hypertrophic cardiomyopathy were excluded. Three of these patients had inadequate pre or post-surgical echocardiographic assessment and are therefore excluded from the study. Other procedures performed and relevant surgical history are documented in Table 1 . (Two of the patients who had previously undergone ventricular septal defect closure had also undergone interrupted aortic arch repair as a neonate, and the VSD was therefore a posterior malalignment defect. However, in both these patients there was also a significant, discrete subaortic membrane). Each compounding factor was assessed to determine whether their occurrence or surgery to alleviate them affected outcome.

Table 1

The incidence of associated lesions and the prior relevant surgical history of the 37 patients

The patients were studied by echocardiography before surgery was undertaken, then at short-term and finally mid-term follow-up.

Echocardiography

LVOT obstruction and AR were assessed on pre and post-operative echocardiograms. Preoperative and early postoperative studies were performed while the patient was in the hospital. Cross-sectional mid-term follow-up was obtained from referring cardiologists in July 1997, and was complete in all but one patient who was lost to follow-up 2 months after hospital discharge.

All patients were examined by two-dimensional and Doppler echocardiography using Acuson 128 XP/5 or 128 XPI1O (Sunnyvale, CA), Hewlett-Packard Sonos 1000,1500 or 2500 (Palo Alto, CA), or Advanced Technologies Laboratory Ultra Mark (Bothwell, WA) ultrasound systems with transducers appropriate for patient size. All studies were recorded on 0.5 ′ super-VHS videotape and were reviewed by two independent observers.

The maximal instantaneous gradient across the left ventricular outflow tract was calculated from the peak spectral Doppler velocity using the modified Bernoulli equation [4],[5]. Aortic regurgitation was characterized as 0, no aortic regurgitation; 1, trivial (small, thin regurgitant jet seen at the valve leaflets, no left ventricular distension or diastolic retrograde aortic flow in the descending aorta); 2, mild (slightly broader jet limited to the LVOT, no ventricular enlargement, minimal or no retrograde flow in the descending aorta, pressure half-time ≧400 ms); 3, moderate (broad regurgitant jet extending into the left ventricle, mild ventricular enlargement, holodiastolic retrograde aortic flow in the descending aorta, pressure half-time 200–400 ms), and 4, severe (wide regurgitant jet extending deep into the left ventricle, marked left ventricular enlargement, holodiastolic retrograde flow in the descending aorta, pressure half-time ≪200 ms). By comparing the pre and post-operative studies it was determined whether our technique reduced the severity of AR.

Surgical approach

All operations were performed on full cardiopulmonary bypass. Moderate hypothermia was used with cooling to 28–33°C for simple cases and 25°C for the more complex procedures. The left ventricle was vented through the left atrium, and after cross-clamping the aorta the heart was arrested using 20 ml/kg of cold oxygenated blood/crystalloid cardioplegia supplemented by cold topical saline. This was repeated at 30-min intervals when the temperature was below 28°C and every 20 min when above this temperature. The aorta was opened by a longitudinal aortotomy which was extended into the non-coronary sinus. Carefully retracting the leaflets of the aortic valve, the subaortic membrane was identified. At this time a very careful inspection of the membrane was performed so that the surgeon identified its full extent, and encroachment of the membrane onto the leaflets of the aortic valve was assessed.

Using a skin hook to gain purchase on the membrane, an incision was made through the white fibrous tissue only to its junction with the muscle. By blunt dissection the membrane was peeled off the muscle all the way round the entire circumference of the subaortic region. The whole extent of the membrane was removed distally down the septum and onto the anterior leaflet of the mitral valve. Attention was then turned to the proximal extension of the membrane and great care was taken to peel it off the undersurface of the aortic valve leaflets, invariably to the middle of the body of the leaflet and often to the free margin. If the membrane broke at any point, another site of origin was prepared and the membrane was removed from this direction. Having completed the membranectomy, the leaflets of the valve themselves were inspected. If they appeared thickened, more tissue was then excised from the ventricular surface of the leaflet until they were thin and compliant.

At this stage the obstruction due to the subvalvar ridge was addressed and an aggressive myomectomy was performed down the length of the interventricular septum, sometimes to the level of the bases of the papillary muscles and also circumferentially around the LVOT onto the free wall of the left ventricle, sparing only the area immediately adjacent to the membranous septum. This was continued until no further obstruction could be seen or felt (when the aortic annulus was of sufficient size to allow this). The ventricle was then washed out with cold saline and the aortotomy closed.

Data analysis

Data are expressed as median (range), mean±standard deviation, or odds ratio with 95% confidence intervals (CI). Wilcoxon's signed-rank test and the Mann–Whitney test were used for comparison of continuous variables between and within groups respectively. McNemar's test was used to compare paired dichotomous variables and Fisher's exact test for unpaired dichotomous variables. Factors that were found to correlate significantly with dichotomous outcome variables by univariate analysis were entered into multivariate analysis using logistic regression with backward stepwise elimination. Non-parametric analysis involving numeric and polychotomous variables was conducted with the Wilcoxon signed-rank test. Spearman's rank and corresponding P-values were calculated for continuous variables.

Results

Patients

There were no operative deaths nor deaths during the period of follow-up. One patient was lost to follow-up, 2 months after hospital discharge. Of the 37 patients, ten had previously undergone resection of a subaortic membrane and were undergoing reoperation for recurrence of severe LVOT obstruction. None of these patients had previously been operated upon in our institute and had not therefore undergone our aggressive surgical approach. None of our patients required reoperation for recurrence of LVOT obstruction. One of our original patients did require reoperation, an aortic valve replacement for moderate AR. This patient had moderate AR preoperatively which improved to mild, following initial surgery. Complications which were experienced during surgery are detailed in Table 2 . Of note, all but one of the cases in which complete heart block requiring permanent pacemaker insertion complicated discrete subaortic stenosis resection were reoperations for recurrence of significant subaortic stenosis. The last patient preoperatively had a right bundle branch block from previous cardiac surgery, at which time a perimembranous ventricular septal defect had been closed. In the three patients in whom an iatrogenic VSD was created, the defect was closed directly using multiple pledgetted sutures in two patients, and using a pericardial patch with interrupted, radially placed, pledgetted sutures in one.

Table 2

Incidence of complications encountered with the aggressive approach to subaortic membrane resection for primary operations and for reoperations

Echocardiography

For the grades of AR, the two observers agreed on 89% of the studies and for those in which there was disagreement, they differed by only one grade.

Preoperative status

The AR grading is shown in Fig. 2 . The mean preoperative LVOT gradient was 66.9±30.4 mmHg (median 64; range 9–138 mmHg). The patient with a gradient of 9 mmHg was referred for mild to moderate AR in association with the membrane. Patients presenting for reoperation had significantly higher peak LVOT gradients (84.3±31.2 mm Hg) than patients undergoing primary subaortic resection (58.3±27.6 mmHg) (P=0.02). The preoperative gradient in patients with mild or moderate preoperative AR (73.9±27.7 mmHg) was greater than in patients with none/trace preoperative AR (57.3±32.0 mmHg), though this was not statistically significance (P=0.09). Age had no significant impact on the results.

Fig. 2

The impact of surgery on the severity of aortic regurgitation early and late post-operatively.

Early post-operative results

Changes in the severity of AR are shown in Fig. 2. Of the two patients who progressed from none/trace to mild AR, one had a bileaflet aortic valve and underwent aortic valve commissurotomy as part of the procedure while the aortic valve of the other was damaged during the initial operation. Overall there was significant improvement in AR following surgery by McNemar's test (p=0.019), though no predictors could be identified.

Factors significantly associated with mild or greater AR in the early post-operative period are summarized in Table 3 . Higher preoperative LVOT gradient (P=0.015) and mild/moderate preoperative AR (P=0.034) were independent predictors of mild or greater AR by multivariate analysis.

Table 3

Factors associated with early and mid-term post-operative aortic regurgitation (mild or greater) following subaortic membrane resection

The mean LVOT gradient at hospital discharge was 15.1±12.2 mmHg (median 12; range 5–60 mmHg), significantly lower than before surgery (P≪0.0001). Early post-operative LVOT gradient correlated significantly with preoperative gradient (r=0.49; P=0.0036) (Fig. 3 ). Patients undergoing reoperation for obstruction had significantly higher early post-operative gradients than patients undergoing primary operation (23.3±17.4 vs. 11.3±80 mmHg; P=0.007).

Fig. 3

Univariate linear regression analysis demonstrating the relationship between early left ventricular outflow tract gradient and preoperative gradient. 'r' and 'P' values are Spearman rank correlation.

Mid-term post-operative results

Follow-up ranged from 2 to 59 months (27.0±16.2 months). One patient was lost to follow-up 2 months after hospital discharge; data from this time were included.

Aortic regurgitation was not significantly different from the early post-operative period (Fig. 2) and this was unrelated to duration of follow-up. One of the patients with late moderate AR underwent mechanical aortic valve replacement 44 months after resection of the subaortic membrane. No factors were identified predicting change in AR during follow-up, nor moderate AR at mid-term. Independent predictors of mild or greater AR at mid-term follow-up were early post-operative aortic regurgitation of a mild or greater degree (P=0.02) and early post-operative outflow gradient (P=0.04) (Table 3).

Left ventricular outflow tract gradient at mid-term follow-up echocardiography was 14.8±12.8 mmHg unchanged from the early post-operative period (P=0.90). Preoperative gradient (P=0.0038; r=0.54) and early post-operative gradient (P≪0.0001; r=0.75) (Fig. 4 ) correlated significantly with follow-up LVOT gradient.

Fig. 4

Univariate linear regression analysis demonstrating the relationship between the late left ventricular outflow tract gradient and the preoperative and early post-operative gradients. Preoperative gradients are shown as squares and early post-operative values as diamonds. 'r' and 'P' values are Spearman rank correlation.

Discussion

Discrete subaortic stenosis is a progressive cardiac abnormality in which the LVOT is obstructed by subvalvar fibromuscular tissue. Though this would appear to be a simple surgical issue, it leads to two particular problems. First, it causes damage to the aortic valve resulting in aortic regurgitation (AR) in up to 50% of patients even with mild gradients. Second, there is a high recurrence rate after surgical resection, between 14 and 27% depending on operative approach. Previous reports show that the incidence of long-term complications is related to the residual gradient after primary operation [6],[7]; minimizing the gradient minimizes later problems. It is particularly important to appreciate the dynamic nature of the LVOT obstruction [8] which may not be evident in the relaxed, cardiopleged heart and take measures to ensure that the obstruction will be completely alleviated when tone returns.

Morphologically 'discrete' subaortic stenosis is a pathological complex, part of which is an endocardial abnormality. This endocardial abnormality involves not only the subaortic ridge but also the leaflets of the adjacent valves [3],[9]. There has been speculation as to the cause of the aortic valvulopathy. Some claim that the leaflet thickening is due to a jet accelerating through the region of obstruction and striking the valve [1],[2]. This would explain the finding that the incidence of AR increases with time [10],[11], though this is not universally accepted [12]. Others suggest that the endocardial abnormality itself is the primary pathological process [3]. In a postmortem study, Feigl et al. [3] showed that a continuous membrane could be seen extending from below the region of obstruction up to, and involving, the leaflets of the aortic valve in 16 of 18 specimens. In seven of these, deformity of the leaflets of the aortic valve could clearly be seen due to the attachment of the accessory tissue producing downward traction on them. Potentially, this could result in both increased LVOT obstruction and AR. In only two of the 18 cases studied was there no involvement of the aortic leaflets by extensions of the subaortic membrane.

Based on this understanding of the morphology and pathophysiology of discrete subaortic stenosis, we hypothesized that if we can recreate normal hemodynamics within the LVOT with no obstruction, by aggressively resecting all obstructing tissue or removing all pathological tissue, the AR associated with the lesion may be improved and the substrate for recurrence of the lesion removed. We therefore adapted an established approach to subaortic stenosis resection by being more aggressive in all aspects of the operation. The conventional surgical approach of sharp dissection of the obstructing ridge with the overlying membrane [13],[14] does not address the more extensive membrane. As the fibrous tissue which fixes and retracts the aortic valve leaflets remains, there can be no improvement in the AR following surgery. MacKay and Ross [8] developed a more aggressive approach of blunt enucleation which involves incising the endocardium over the obstructing ridge and peeling it off the underlying myocardium. Using this technique the 'finger-like projections of tissue' that creep onto the aortic valve can be removed intact with the ring of fibrous tissue. A myotomy or myomectomy was then performed to relieve the obstruction. We adapted this technique in two ways. First, we were meticulous in removing all pathological tissue from the valve leaflets, the sub-commissural trigones and the left ventricular outflow tract. Second, we performed a very aggressive circumferential (if necessary) myomectomy to relieve the obstruction.

This aggressive approach appears to be effective in reducing both the severity of AR and the LVOT gradient, at least at short- and mid-term follow-up. Twelve of the patients showed improvement in the severity of AR and only two showed worsening at early follow-up. Particularly of note, one of these patients underwent direct aortic valve surgery while the other suffered damage to the valve leaflets during the procedure. Mid-term follow-up of our patients suggests that the improvement in AR seen immediately after surgery persists. As shown in previous studies, by logistic regression analysis early post-operative gradient was the strongest predictor of mild or greater AR at mid-term follow-up; duration of follow-up and presence of mild or greater AR preoperatively, though also independent predictors, were weaker. Radical excision of all pathological tissue achieving minimal early post-operative gradient may therefore reduce the incidence of late AR.

Similarly the LVOT gradient was significantly reduced at early follow-up and this improvement was maintained at mid-term follow-up. As early post-operative gradient correlated significantly with mid-term post-operative gradient, a good hemodynamic result from surgery allows the greatest chance of relief of obstruction long-term. However, patients presenting for reoperation had higher gradients at both follow-up time points than patients undergoing primary surgery without a significant difference in gradient between the two time points. This suggests that reoperation is less effective than primary operation in relieving LVOT obstruction even though the progress of the obstruction can be halted.

Compared with previous studies the incidence of complications appears to be higher in our series than previously reported [12],[15],[16],[17],[18],[19], particularly permanent complete heart block requiring a permanent pacemaker. However, of our patients who suffered this complication, four out of five (80%) were patients undergoing reoperation for recurrent obstruction; the incidence of complete heart block in patients undergoing initial surgery was only 3.7% (1/27). Further, this last patient had previously undergone closure of a ventricular septal defect and suffered blockade of the right bundle branch during this procedure. The incidence of complete heart block undoubtedly reflects the very aggressive approach which we adopt and the fact that we perform a circumferential myomectomy to remove all the obstruction. The LVOT gradient in patients undergoing reoperation was significantly higher and therefore even more aggressive muscle resection was undertaken which may account for the increased incidence of complete heart block. We would therefore suggest that our aggressive approach is justified particularly in first time operations in an attempt to prevent the need for later reoperation, and that it may be because a less aggressive approach was adopted initially that the patients required repeat operation. Though the duration of follow-up is relatively short, the reoperation rate in our series to date is 0% which would support our contention, but we must wait for the long-term results before our approach can be completely justified.

In conclusion, our results show that extended resection of the subvalvar membrane and thinning of the aortic valve leaflets combined with aggressive myomectomy produces excellent relief of the obstructive element of subaortic stenosis and frees the valve leaflets, thereby significantly reducing the associated aortic regurgitation. At primary operation this is associated with a low complication rate. Reoperation striving to achieve similar hemodynamic results is associated with a significantly higher rate of complications. We contend that an aggressive approach to normalize LVOT hemodynamics at the first operation in order to avoid later reoperation can therefore be justified. Late follow-up is crucial to confirm this conclusion.