Sternotomy Approach for the Modified Blalock-Taussig Shunt
Background Since 1990, sternotomy has been the preferred approach for construction of a modified Blalock-Taussig shunt (MBTS) at Children’s Hospital, Boston, Mass. In retrospect, we sought to test the hypothesis that this approach yields less mortality and morbidity than the traditional thoracotomy approach.
Methods and Results One hundred four primary MBTSs with polytetrafluoroethylene grafts were constructed in patients from January 1988 through December 1992. Fifty-two shunts were constructed by thoracotomy approach and 52 by sternotomy approach. Fifteen of the thoracotomy patients were less than one month of age (8 less than 7 days), while 36 of the sternotomy patients were less than 1 month of age (20 less than 7 days). There were 10 shunt failures and 3 hospital deaths in the thoractomy group and 4 shunt failures with 6 hospital deaths in the sternotomy group. The overall hospital mortality rate for the group was 8.7% (9 of 104). The operative route was not a significant predictor of hospital mortality (P=.30). However, there was a significant difference between the two operative approaches in shunt failure, with shunts that were created by thoracotomy four times more likely to fail than those created by the sternotomy route (odds ratio, OR, 3.88; 95% CI, 1.01 to 15.03; P=.049). The side of the shunt was also a significant predictor of failure with left-side MBTSs four times more prone to failure (OR, 4.02; 95% CI, 1.19 to 15.25; P=.025).
Conclusions The sternotomy route is technically less challenging and is associated with fewer shunt failures than the classic thoracotomy approach. The potential theoretical disadvantages of this method for future sternal reentry for subsequent procedures was not apparent but requires prospective analysis.
In the present era, primary correction is the preferred approach to the neonate or young infant with a cardiac anomaly who has two ventricles. However, when only one functional ventricle is present or pulmonary blood flow is reduced, an initial palliative systemic-to-pulmonary arterial shunt is mandatory. The shunt should not supply excessive pulmonary blood flow that might result in elevated pulmonary vascular resistance or impair ventricular performance secondary to excessive volume load. The shunt should not distort the pulmonary arteries, particularly distally, and it should encourage uniform growth and development of the pulmonary arteries. Of the numerous options for connection between the systemic and pulmonary circulations that have been described,1 2 3 4 5 6 the shunt that has generally been accepted as most likely to fulfill these criteria is the modification of the Blalock-Taussig shunt in which a PTFE tube graft is inserted between the subclavian and pulmonary arteries.7 8 9 10 11 12 13 The preferred surgical approach to this shunt when it was first described by de Leval7 was through a left thoracotomy with anastomosis to the left PA. However, in a child with a single ventricle, it is preferable to perform the anastomosis on the side of the superior vena cava, ie, most commonly on the right, for later conversion to a cavopulmonary shunt. In this way, any distortion of the PA that results from the distal anastomosis can be readily incorporated in the cavopulmonary anastomosis of a Fontan procedure with or without a preliminary bidirectional cavopulmonary anastomosis. Furthermore, takedown of a LMBTS, working from the midline, carries significant risk of injury to the left phrenic nerve.
From 1984 to 1989, our usual approach for construction of the RMBTS was a right thoracotomy. During this period, a number of disadvantages of this approach became apparent. The dissection of the bifurcation of the right PA and subsequent placement of Silastic vessel loops for control during the anastomosis resulted in an important incidence of distortion of the right PA just beyond the upper lobe takeoff that was difficult to repair at subsequent procedures. Because the anastomosis was relatively distal on the right PA, considerably more flow was directed into the right than the left PA, resulting in unbalanced development of the PAs. The relatively distal dissection of the subclavian artery at the apex of the chest was complicated occasionally by Horner’s syndrome. Surgical trainees often found performance of the proximal anastomosis challenging. Thoracotomy wounds in neonates who remained cyanotic because of the nature of their anomalies were more often complicated by delayed healing than were sternotomy incisions. The development of late scoliosis has been reported after thoracotomy incisions. We also have noted the development of chest wall–to-lung collaterals in patients with previous thoracotomies. Finally, there is a cosmetic disadvantage to use of two incisions rather than a single sternotomy incision.
Experience with construction of the MBTS through a sternotomy as part of first-stage palliation of hypoplastic left-heart syndrome or for pulmonary atresia with intact ventricular septum suggested that this was a preferable approach for the shunt procedure alone.14
We wished to test the hypothesis that the median sternotomy approach for construction of the MBTS increased neither morbidity or mortality. We retrospectively review our consecutive experience with the MBTS from January 1, 1988 through December 31, 1992.
During the period from January 1, 1988, through December 31, 1992, 104 patients with complex cyanotic congenital heart lesions underwent primary MBTS operations at Children’s Hospital, Boston, Mass (Table 1⇓). By chance, there were 52 sternotomy and 52 thoracotomy patients (61 boys and 43 girls). The age of all patients was 232.8±87.4 days (mean±SEM) (range, 1 day to 21.5 years). The median ages of the thoracotomy and sternotomy groups were 84 days (range, 1 day to 21.5 years) and 14 days (range, 1 day to 410 days), respectively. Forty-nine percent of all patients undergoing primary shunting operations were neonates, including 69% from the sternotomy and 29% from the thoracotomy groups (Table 2⇓).
Inclusion and Exclusion Criteria
Only patients undergoing primary MBTS were included in this retrospective analysis. Excluded from this study were all patients with hypoplastic left-heart syndrome undergoing the staged Norwood operation, babies with pulmonary atresia who underwent a simultaneous valvotomy and/or right ventricular outflow tract augmentation, and any other patients undergoing associated intracardiac procedures in addition to a systemic-to–pulmonary arterial shunt. Twenty-six babies with transposition of the great arteries who underwent preliminary shunting and PA banding before the arterial switch were included in the groups. Of the shunts received by 26 infants as part of a rapid two-stage arterial switch, 16 were placed by thoracotomy and 10 by sternotomy. Several children had various shunts performed elsewhere before coming to our institution, and this factor influenced the choice and location of future shunts. Therefore, these infants were not included in our review.
Operative Technique for Median Sternotomy Approach
The patient is carefully positioned with the neck extended by a shoulder roll. A median sternotomy extending a few millimeters above the sternal notch is used for entry. The thymus is excised, leaving behind a small cervical remnant. The pericardium is opened and gently suspended on the right side with traction sutures. The innominate artery and right subclavian artery are dissected free of periadventitial tissue. The right PA is dissected out between the superior vena cava and aorta. A PTFE graft of suitable diameter is selected (usually 3.5 mm for the average neonate), and the proximal end is beveled (Fig 1⇓). Heparin is not routinely administered. A C-clamp is applied to the distal innominate and proximal subclavian arteries. An arteriotomy is made on the inferior aspect and an anastomosis created between the PTFE graft and the distal innominate/proximal subclavian arterial junction with a continuous monofilament suture (polypropylene or PTFE). A second clamp is then applied distally across the graft, permitting removal of the C-clamp from the innominate and subclavian arteries. The graft is then passed underneath the innominate vein and cut to final length. A C-clamp is applied to the right PA ensuring that flow into the left PA from the ductus or main PA is not compromised. The PA is opened longitudinally and the distal anastomosis between the PTFE graft and the PA completed. The clamp is released to allow back-bleeding from the PA. Hemostasis is ascertained and antegrade flow is permitted by releasing the clamp on the graft. An open shunt should result in a drop in systolic and diastolic blood pressure and an increase in arterial oxygen saturation. A right atrial line is usually placed through a purse-string suture in the right atrial appendage to provide central access, with minimal risk to the large central veins. A patent ductus arteriosus, if present, may be ligated. A single chest tube is placed extrapleurally through a separate right lateral stab-wound incision to drain the superior mediastinum. The sternum is closed in standard fashion.
The criteria used to define early shunt failure were (1) complete occlusion during the hospitalization period, (2) the need to return to the operating room for a second shunt, or (3) complete repair during the same hospitalization.
After initial hospital discharge, late shunt failure was assigned to patients presenting with complete occlusion, progressive desaturation, or polycythemia that led investigators to document important stenosis or occlusion.
Distortion of the PA related to the distal shunt anastomosis was defined at angiography by the cardiologist or at subsequent operative interventions that required concomitant pulmonary arterioplasty.
Cellulitis of the skin requiring antibiotics, wound breakdown or suppuration, and mediastinitis constituted the definition of a wound infection in this series.
The risk of shunt failure was compared between the two operative groups. Logistic regression was performed to estimate the relative risk of this primary outcome and other morbid end points. For each OR, the 95% CI and two-sided probability values were calculated. Woolf’s method was used with the variance approximated by a Taylor series expansion.15 All statistical analyses were performed using SPSS computer software (SPSS Inc).16 A two-tailed probability based on Fisher’s exact test was used to compare proportions between the two groups regarding complications.
The mean diameters of the conduit material were 4.10±0.52 and 3.66±0.31 mm for the thoracotomy and sternotomy groups, respectively (t=5.21, df=102, P<.001) (Fig 2⇓).
There were 9 total hospital deaths, ie, a hospital mortality rate of 8.7%. In the sternotomy cohort of 52, none of the 6 deaths was directly attributable to shunt malfunction, although excessive pulmonary blood flow could have been contributory (Table 3⇓). In the thoracotomy group of 52, 1 of the 3 deaths was directly attributable to shunt occlusion (Table 4⇓). By univariate analysis, the operative approach was not a significant predictor of hospital mortality (P=.30).
The operative approach was a significant predictor of shunt failure. Nineteen percent (10 of 52) of the thoracotomy group experienced early or late shunt failure (Table 5⇓), whereas only 7.7% (4 of 52) of the sternotomy group had a similar fate. Thoracotomy patients were four times more likely to experience shunt failure than their sternotomy counterparts (OR, 3.38; 95% CI, 1.01 to 15.03; P=.049).
RMBTS Versus LMBTS
The side of the MBTS (RMBTS=82, LMBTS=22) was predictive of shunt failure, with LMBTS performing more poorly (OR, 4.02; 95% CI, 1.19 to 13.57; P=.02). Table 6⇓ presents the summary statistics for the ORs and logistic regression with respect to hospital mortality and shunt failure.
Multivariate risk factors for mortality included the side of shunt (P=.03), sex (P=.05), and graft size (P<.01). Multivariate risk factors for shunt failure included operative approach (P=.04), side of shunt (P=.02), and sex (P<.01). In short, patients with an LMBTS, male sex, and/or smaller graft size had poorer survival rates. Patients with an LMBTS, male sex, and treated by the thoracotomy approach were more likely to have shunt failure.
Two patients were documented to have had Horner’s syndrome. Both of these patients presented in infancy with transposition of the great arteries and low left ventricular pressure and underwent the rapid two-stage arterial switch procedure. They represented 8% (2 of 26) of all infants undergoing the rapid two-stage arterial switch procedure during this 5-year period. These two infants were among 16 in this group who had preparatory PA banding and systemic-to–pulmonary arterial shunting by thoracotomy. There was no incidence of Horner’s syndrome among the remaining 10 infants who had the rapid two-stage procedure using the sternotomy route or any incidence of Horner’s syndrome in the sternotomy group as a whole.
Recurrent Laryngeal Nerve Injury
There was one patient with recurrent laryngeal nerve injury in this series, in the sternotomy group (1 of 104).
Phrenic Nerve Injury
One patient in each group had phrenic nerve injury, for an overall frequency of 2% (2 of 104).
Four percent (2 of 52) of the thoracotomy group had an incisional wound infection compared with 8% (4 of 52) of the sternotomy group (P=.68). These infections were all superficial.
Six percent (3 of 52) of the sternotomy group developed a pericardial effusion requiring drainage. There was no such incidence in the thoracotomy cohort (P=.24).
Six percent (3 of 52) of the sternotomy patients developed an important pleural collection that required tube thoracostomy or drainage. Two percent (1 of 52) of the thoracotomy group developed this complication (P=.27).
There were no cases of chylothorax in this series.
There were 4 cases (thoracotomy=1, sternotomy=3) of pulmonary overcirculation with decreased systemic blood flow. This occurrence required resternotomy for ligation of a patent ductus arteriosus and/or narrowing of the shunt.
Of the 46 hospital survivors undergoing initial shunting procedures through a midline approach, 40 (87%) underwent resternotomy. The in-hospital mortality rate of the resternotomy group was 13% (5 of 40).
In contrast, of the 49 in-hospital survivors undergoing thoracotomy for MBTS, 38 (78%) progressed to more definitive surgery with a primary sternotomy. The in-hospital mortality rate of this reoperative group was 8% (3 of 38) (P=.71).
Of the 78% of the thoracotomy group continuing to further definitive surgery, 16% (6 of 38) required a concomitant pulmonary arterioplasty. In contrast, of the 87% of the sternotomy group undergoing resternotomy, 10% (4 of 40) required an accompanying pulmonary arterioplasty (P=.74).
Interestingly, sex emerged as a predictor of shunt failure. In the male population, thoracotomy had a significantly higher risk of shunt failure (OR, 7.33; 95% CI, 1.41 to 38.25; P=.02). In the female population, neither operative approach (P=.82) nor side of shunt (P=.85) was a significant predictor of shunt failure.
Age was not a significant predictor of shunt failure (P=.051). Operative approach and the side of shunt creation were not significant predictors of nerve palsies (phrenic, recurrent laryngeal, or dorsal stellate ganglion), wound infections, or pleural or pericardial effusions.
The MBTS is traditionally performed through a thoracotomy approach. There are several disadvantages of this approach. This review documents a significantly higher rate of shunt failure in patients undergoing thoracotomy in our institution. The more distal dissection along the right pulmonary arterial tree probably results in distal PA distortion, particularly, beyond the takeoff of the right upper lobe. In contrast, a median sternotomy approach limits the dissection to the right PA proximal to the right upper lobe takeoff. Any potential distortion from the distal anastomosis can be easily incorporated into a bidirectional Glenn shunt or Fontan anastomosis. Since the shunt is placed more centrally on the pulmonary arterial tree, flow distribution to the right and left lung, barring nonconfluence or distal stenoses, is more uniform. Amato and colleagues,6 who used the sternotomy approach for central systemic-PA shunts, reported the absence of pulmonary vessel distortion in patients undergoing a central PTFE shunt compared with other patients who received classic Blalock-Taussig shunts and MBTS. Furthermore, they observed more preferential flow to the side of construction and subsequent uneven growth of the branch pulmonary arteries in those patients undergoing classic Blalock-Taussig shunts and MBTS.
In this series, although the number of nerve palsies was small and there was no significant difference between the two groups, the only two patients developing Horner’s syndrome underwent an initial thoracotomy as part of a rapid two-stage procedure toward arterial switching for D–transposition of the great arteries. Horner’s syndrome is well described as an important complication of the thoracotomy approach to MBTS. Although the right recurrent laryngeal nerve is at greater risk with the thoracotomy approach because of the more distal dissection of the right subclavian artery, we did not observe a significant difference in incidence between these groups.
Analysis of mode of death of the patients undergoing MBTS by sternotomy suggests that excessive shunt flow may have contributed to a low cardiac output and therefore increased the risk of cardiac arrest and death. We learned early in our experience that because the proximal anastomosis of the shunt is placed more proximal on the innominate/subclavian arterial system and because the proximal anastomosis is technically better exposed and therefore less likely to be narrowed by suture technique, it is important to use a 3.5-mm tube graft for the average-sized neonate rather than a 4-mm graft. A 4-mm–tube graft remains the diameter of choice if a thoracotomy approach is used for shunt placement. Other considerations in determining the size of the graft selected include the presence of an additional source of pulmonary blood flow (pulmonary stenosis versus pulmonary atresia) and an estimate of pulmonary vascular resistance.
Although the sternotomy route for MBTS makes for a technically less demanding operation and has the cosmetic advantage of a single sternotomy incision, the need to undertake a resternotomy for subsequent bidirectional cavopulmonary anastomosis or Fontan procedure raises a theoretical disadvantage. The problems of injury to the heart or great vessels during reentry and prolonged operative time at subsequent reoperation were not observed. Evaluation of the patients in this series who went on to definitive repair revealed no important disadvantage. There was no important difference in the subsequent hospital mortality.
In our series, the LMBTS was four times more prone to failure than the contralateral RMBTS. This occurrence appears to be related to the complexity of cardiac disease, with anatomic constraints forcing a decision to approach the shunt from the left. In addition, the left side is used much less frequently and is perhaps therefore more prone to technical misadventure.
The finding that male patients have a significantly higher risk of shunt failure among the thoracotomy patients is curious. Male sex in this model may be associated with some other surrogate variable (eg, weight or complexity of the heart lesion) as a predictor for shunt failure.
In conclusion, we advocate the sternotomy approach to MBTS in neonates and infants. This approach is technically easier to perform, is cosmetically preferable, and perhaps is hemodynamically superior. The in-hospital mortality rate is acceptable, and this route is associated with less shunt failure. Correction of any pulmonary distortion is easily incorporated at subsequent procedures. Our experience with reoperative cardiac surgery suggests that the advantages of the sternotomy approach are not outweighed by the disadvantages of a subsequent resternotomy.
Selected Abbreviations and Acronyms
|LMBTS||=||left-side modified Blalock-Taussig shunt|
|MBTS||=||modified Blalock-Taussig shunt|
|RMBTS||=||right-side modified Blalock-Taussig shunt|
Reprint requests to Richard A. Jonas, MD, Department of Cardiac Surgery, Children’s Hospital, 300 Longwood Ave, Boston, MA 02115.
- Copyright © 1995 by American Heart Association
Kay PH, Capuani A, Franks R, Lincoln C. Experience with the modified Blalock-Taussig operation using polytetrafluoroethylene (Impra) grafts. Br Heart J. 1983;49:359-363.
Moulton AL, Brenner JI, Ringel R, Nordenberg A, Berman MA, Sohaila A, Burns J. Classic versus modified Blalock-Taussig shunts in neonates and infants. Circulation. 1985;72(suppl II):II-35-II-44.
Laks H, Fagan L, Barner HB, Willman VL. The Blalock-Taussig shunt in the neonate. Ann Thorac Surg. 1978;25:220-224.
Schlesselman JJ. Case control studies: design, conduct, analysis. New York, NY: Oxford University Press; 1982.
Norusis M. SPSS for Windows, Release 6.0. Chicago, Ill: SPSS Inc; 1993.
Hosmer D, Lemeshow S. Applied Logistic Regression. New York, NY: John Wiley & Sons; 1993.