Introduction
In young patients with diminished pulmonary blood flow, a patent ductus arteriosus (PDA) is needed to maintain stable hemodynamics. Such a condition often occurs in neonates with complex congenital heart disease whose hemodynamic stability depends on a PDA. These cardiac lesions are called duct-dependent congenital heart defects [
1].
The conventional emergency treatment to maintain pulmonary blood flow is prostaglandin E infusion [
2]. However, this treatment can be administered only by intravenous infusion, hence the impracticality of keeping neonates with duct-dependent pulmonary circulation in hospital until the time of definitive surgery [
3]. Then, a surgical shunt called the Blalock–Taussig shunt (BTS), which is a direct end-to-side anastomosis of the subclavian artery to the ipsilateral pulmonary artery, is introduced and later modified by the interposition of a tube graft (modified BTS) [
4]. Over time, other different surgical systemic-pulmonary artery shunts have been proposed [
5]. Despite its widespread use and technical improvements, surgical shunt has been reported to be associated with significant mortality and morbidity [
6,
7].
With minimally invasive transcatheter approaches, ductus stent implantation has long been proposed as an effective alternative to surgical systemic-pulmonary artery shunt in patients with duct-dependent pulmonary circulation [
8‐
10]. However, complications such as worsening cyanosis, bleeding, vessel rupture, arterial duct spasm or acute stent thrombosis have also occurred after stenting [
11].
Therefore, the safety and efficacy of arterial duct stent and surgically created shunts in patients with duct-dependent pulmonary circulation are still controversial. Consequently, we performed this meta-analysis to compare the outcomes of the two approaches in an attempt to support evidence for clinical strategies.
Discussion
To our knowledge, this is the first meta-analysis to compare ductus stent implantation and surgically created systemic-pulmonary shunt in patients with duct-dependent pulmonary circulation. We found that procedural complications, ICU and hospital stay, and total mortality favored the stent group. The proportion of patients with a single ventricular or double ventricle was significantly different between the two groups. Additionally, other outcomes showed no significant differences between the two groups.
Duct-dependent congenital heart defects involve single or double ventricle physiology with kinds of cardiac anomalies, such as pulmonary atresia with intact ventricular septum or ventricular septal defect, tetralogy of Fallot, pulmonary stenosis, tricuspid atresia, transposition of great vessels, Ebstein anomaly, etc.[
25]. Even though not all included studies confirmed that there was a tendency in patient selection [
17,
18,
21], the proportion of patients with a single ventricle was higher in the shunt group, and the proportion of patients with a double ventricle was higher in the stent group [
16,
20].
Many duct-dependent congenital heart defects either require staged palliation or can be corrected only at a later age. In addition, in the current era, BTS as palliation is a procedure that is almost exclusively performed in the neonatal period or early infancy [
26]. However, there are still concerns about the significant morbidity and mortality after BTS [
27,
28]. Ductus stent implantation, although not widely accepted, has the advantage of being minimally invasive, which avoids a median sternotomy or lateral thoracotomy and exposure to cardiopulmonary bypass [
7]. However, in some cases with bizarre, long and tortuous PDA, it still represents a major technical challenge, which could lead to procedural failure and pulmonary artery distortion; thus, we think it is inappropriate to implant the stent [
25,
27].
The procedure-related complications of BTS involve thrombosis, pleural effusion, chylothorax, phrenic and vagal nerve palsy, distortion, and distal pulmonary artery stenosis [
5]. In addition, the complications of stent implantation involve thrombosis, embolism, ductal spasm, migration of stent, and branch pulmonary artery stenosis [
25,
26]. Amoozgar et al. (0 vs. 30%; p < 0.05) and Mallula et al. (7.7% vs. 37.5%; p < 0.05) reported a lower incidence of complications after the stent implantation than the surgical shunt [
17,
19]. However, McMullan et al. (0 vs. 7%; p = 1.0) and Glatz et al. (13.2% vs. 21.5%; p = 0.07) reported no significant differences in the incidence of complications between the two groups [
18,
20].
Some cases with complications such as thrombosis embolism, migration of stent, stent stenosis, or branch pulmonary stenosis require unplanned reintervention [
1,
29]. In addition, intimal proliferation at the implantation site is almost universal in the first 3–6 months, requiring planned reintervention in the majority of patients [
10,
30]. Although stent implantation could potentially aggravate branch pulmonary artery stenosis, with standard initial palliation with a surgical shunt, pulmonary arterioplasty is frequently a part of surgical repair, and postoperative branch pulmonary stenosis requiring reintervention is a common late problem in most pulmonary atresia patients [
27].
Amoozgar et al. reported the absence of reintervention in the stent and shunt groups including 35 patients [
17]. In addition, the incidence of reintervention was similar (stent vs. shunt: 25% vs. 26%) in both groups by McMullan et al. [
18]. Mallula et al. (58.3% vs. 14.3%, p < 0.05) and Bentham et al. (39.8% vs. 24.0%, p < 0.05) reported increased reintervention in the stent group [
19,
21]. Only Glatz et al. (11.3% vs. 20.7%, p < 0.05) reported decreased intervention in the stent group [
20]. When combined with more frequent planned reintervention in another study, the overall reintervention (48.6% vs. 2.2%, p < 0.05) was more common in the stent group [
20,
21].
Due to the minimal invasion and lower complications, patients receiving stent implantation are supposed to recover faster than those receiving surgical shunt. Three studies reported postoperative ventilation time, all with a median time of 1 day in the stent group, which was less than 3 or 4 days in the shunt group [
19‐
21]. Relatively, the length of ICU stay and hospital stay were both shorter according to the studies by Glatz et al. and Bentham et al. [
20,
21]. Furthermore, Goldtein et al. compared the costs for hospitalization and the first year of life between the two groups and found that there were lower costs for patients who received stent implantation [
22].
With respect to mortality, there were no significant differences in early mortality between the two groups reported by each study [
16‐
19]. However, Bentham et al. reported a reduced risk of total death at follow-up (hazard ratio 0.25; 95% CI 0.07–0.85; p < 0.05) in the stent group when compared with the shunt group [
21], and Glatz et al. reported no significant differences in the risk of total death (hazard ratio 0.64; 95% CI 0.28–1.47; p = 0.29) between the two groups [
20]. Considering risk factors identified for mortality such as preoperative mechanical ventilation and underlying cardiac anatomy, Glatz et al. and Bentham et al. had constructed propensity-adjusted models with hope to avoid the effects of confounders [
20,
21]. Although procedural mortality has significantly declined over time, it is still measurable and has driven the consideration of alternative approaches [
28]. With decreased total mortality, ductus stent implantation seems to be a preferable alternative.
Since most patients underwent follow-up cardiac catheterization before surgical repair in less than 1 year, the follow-up period before the next stage repair of most studies is no more than 1 year, as shown in Table
1 [
16‐
18,
21]. Santoro et al. (210 ± 90 vs. 360 + 120; p < 0.05) and Mallula et al. (99 ± 67 vs. 131 ± 57; p < 0.05) reported that patients in the stent group waited for less time to definitive repair than the shunt group [
16,
19]. However, McMullan et al. (153 ± 136 vs. 196 ± 91; p = 0.20) and Bentham et al. (231 ± 40 vs. 243 ± 32; p = 0.56) reported there were no significant differences in the interval to next stage repair between the two groups [
18,
21]. Only Glatz et al. (178 ± 25 vs. 150 ± 15; p < 0.05) reported less waiting time in the shunt group [
20].
Both ductus stent and surgical shunt could maintain blood flow to the lungs, thereby promoting the growth of pulmonary arteries [
30]. However, the comparisons of the stent and shunt on the growth of pulmonary arteries remain controversial [
5]. During the follow-up, our synthetic results showed that there were no significant differences in the growth of pulmonary arteries, except for the left pulmonary artery, which was larger in the shunt group. This finding might be because the surgical shunt produced both the overgrowth of the contralateral pulmonary artery and a lesser development of the ipsilateral pulmonary artery compared with the percutaneous approach, presumably due to unfavorable graft geometry and flow direction to the pulmonary vascular bed [
16]. Thus, blood flow through the ductus stent with an optimal angle enters the pulmonary arteries centrally, with the potential for relatively symmetrical flow to each branch pulmonary artery [
20]. Stent implantation might provide a more evenly distributed pulmonary blood flow and promote more balanced growth of the pulmonary arteries [
16,
27].
After the sensitivity analysis, we found that the study by Bentham et al. [
21] was the source of heterogeneity in the pooled estimates of hospital stay. Additionally, we excluded this study and found there was no directional change in the pooled estimates, which suggested that the result was stable. According to Papachristofi et al., many factors, such as individual patient risk, center, surgeon and anesthetist, could have an effect on the length of hospital stay after cardiac surgery [
31]. This finding suggested that these factors might affect the hospital stay in the study by Bentham et al. [
21], which made this study a source of heterogeneity.
Limitations
Several limitations exist in this study. First, the number of studies and the sample size are limited due to the comprehensive procedures performed. The results should be drawn with caution due to the generally limited number of studies, small sample size, and heterogeneity in some analyses. Meanwhile, the synthetic analysis of some of the secondary outcomes was conducted in only 2 studies. According to Cochrane Handbook, even though the meta-analysis could be conducted with more than 2 studies, the limited number of included studies could downgrade the quality of evidence level, especially increasing the risk of publication bias [
15]. Hence, we did not only report the pooled estimates of some secondary outcomes but also focused on the individual results of those original studies. Second, all included studies were nonrandomized studies. Observational analyses of this nature fail to fully account for selection bias subtly and inadvertently introduced into the study, which cannot be controlled. Third, since original studies did not compare the outcomes between the stent and shunt among the single ventricle and double ventricle subgroups, nor did they only include patients with a single or double ventricle, we were unable to make the subgroup analysis stratified by ventricle physiology, although it may be a potential source of the heterogeneity. Finally, because some limitations could not be overcome, randomized controlled trials with more data and longer follow-up durations are needed to confirm our findings.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.