A meta-analysis of perventricular device closure of doubly committed subarterial ventricular septal defects

Ventricular septal defects (VSDs) account for 20% of all forms of congenital heart defects, and approximately 5–7% of VSD cases are doubly committed subarterial VSDs (dcsVSDs) in Asian populations, which have a low tendency for spontaneous closure and a high incidence of aortic valve prolapse [1‐4]. Early intervention is recommended for patients with dcsVSDs. Surgical repair of dcsVSDs under cardiopulmonary bypass (CPB) is considered the gold standard treatment. However, this procedure cannot avoid the potential risk of CPB-related complications, the need for a blood transfusion, and the presence of a surgical incision scar or prolonged recovery [5, 6]. With the improvement of technology and development of devices, transfemoral device closure of dcsVSDs has also been applied in some cardiac centers [7, 8]. However, such an approach is still limited by technical difficulty, caused by the special location of dcsVSDs, which results in longer radiation times and lower success rates. In recent years, perventricular device closure of dcsVSDs has been developed as an alternative to conventional surgical repair in China [9‐17]. No meta-analysis focusing on perventricular device closure of dcsVSDs has been reported. This study aimed to obtain pooled estimates of the success and morbidity rates after perventricular device closure of dcsVSDs, based on a meta-analysis of the current literature. The results might further guide research on the risk factors for complications to achieve better outcomes with fewer complications.

Perventricular device closure of perimembranous VSDs (pmVSDs) and muscular VSDs (mVSDs) has been confirmed to be safe and effective [20, 21]. Recently, this technology has also been used in patients with dcsVSDs. However, due to the special location of dcsVSDs, the safety and efficacy of perventricular device closure of dcsVSDs is still unclear. In this systematic review, we have attempted to evaluate the efficacy and safety of this technology.

The included studies were 5 case series and 4 case-control studies of high quality. The invasive procedure limited the blinding of the participants and resulted in a lack of RCTs. We attributed the funnel asymmetry to publication bias. Studies with promising results had an increased likelihood of being accepted and published. Thus, publication bias may contribute to a higher pooled success rate. Fortunately, the trim-and-fill test did not show any trimming or filling, and the results were stable, which suggested that the bias was acceptable.

We defined operational success as patients without fatal or severe early-term or late-term complications. The pooled success rate was 0.89 (95% CI: 0.86–0.93, I² = 26.5%, P = 0.208) for 9 studies with 459 patients. Further meta-regression analysis indicated no significant correlation between the success rate and the following factors: publication year, sample size, study type, mean age, mean weight, mean VSD size, and ratio of device size/weight, which may indicate the short learning curve and promotability of this technology. Compared with surgical repair, perventricular device closure does not require CPB. Compared with the transfemoral approach, the perventricular approach provides direct access and facilitates manipulation of the device position and orientation during device deployment. We attributed this to the shorter delivery path. A shorter delivery path also minimizes the risk of intracardiac structural damage due to catheter friction or rubbing. Thus, for experienced cardiac surgeons, the learning curve is short, and the promising prospects of this technology are easily promoted.

Only patients with isolated dcsVSDs were included, and patients with other coexisting cardiac anomalies requiring surgical intervention, severe pulmonary hypertension, or significant aortic prolapse and VSDs larger than 10 mm were excluded. However, no uniform patient inclusion criteria were applied among the cardiac centers. Whether patients with VSD sizes between 5 mm and 10 mm or patients with mild aortic valve prolapse (AVP) could undergo this procedure is unclear. Most studies recommended that the VSD size should be less than 10 mm. However, Cao and his colleagues recommended that the VSD size should be less than 5 mm. They found that in cases with a VSD size > 5 mm, the effective contact area of the occluder became small and, in cases with a VSD size < 5 mm, there was a relatively small opening in the superior margin. Thus, in cases with a VSD size < 5 mm, the device rarely affects the pulmonary valve and is less likely to be displaced from the original position [16]. The incidence of occluder displacement in patients with VSDs > 5 mm was greater than the incidence of displacement in patients with VSDs < 5 mm. Hu and his colleagues also reported that a dcsVSD size ≥5 mm was a predictor of percardiac device closure failure (odds ratio, 41.25; 95% confidence interval, 4.69–362.72; P < 0.001) [12]. Meta-regression analysis showed no correlation between the success rate and mean VSD size. Patients with VSD sizes between 5 mm and 10 mm may still be suitable for this procedure. Most studies did not exclude patients with AVP (below severe). However, multivariate logistic regression performed by Zhang and his colleagues showed that procedure failure was associated with the occurrence of preoperative AVP (even in mild degree). The AVP may lead to underestimation of the VSD size and increase the risk of device dislocation. The interface between the aortic valve and device may cause procedure failure as well as procedure-induced valve complications. Unfortunately, most enrolled patients did not provide enough information to conduct further analyses [15].

The pooled rate of severe intraoperative complications was 0.106 (95% CI: 0.073–0.140, I² = 70.7%, P = 0.208). A total of 53 patients were converted to conventional surgical repair, including 22 patients with new mild to significant aortic regurgitation, 13 patients with significant residual shunting, 8 patients with device dislocation, and 3 patients with failure to establish a path. New mild to significant aortic regurgitation and significant residual shunting were the most common reasons for conversion. The pooled rates of failure to establish a path, device dislocation and severe arrhythmias were low in perventricular device closure of dcsVSDs. The reason for the low incidence of failure to establish a path was that such a procedure could provide a perpendicular approach via the right ventricular surface toward the dcsVSD. Suitable puncture was determined by depressing the right ventricular free wall with an index finger to find the strongest tremor site under transesophageal echocardiography (TEE) guidance. Most complications disappeared after removal of the device, which suggested the importance of choosing a suitable device size. The asymmetrical occluder was the most widely used occluder in these studies. The device size was selected according to the TEE measurement and allowed a margin of 0 to 2 mm in excess of the diameter of the VSD. Suitable device size is a key factor in preventing procedure-related AR. Lin and his colleagues reported that the ratio of the device diameter to weight and procedure-related AR had a statistical association (OR = 4158.325, 95% CI 4.388–3,941,113.209, P = 0.017) [10]. Although the pooled rate of device dislocation was only 0.000 (95% CI: − 0.001-0.001, I2 = 16.4%, P = 0.301), TEE was necessary to assess the presence of the device position for 10–20 min after device placement [18].

The pooled rate of severe postoperative complications was promising, only 0.000 (95% CI: − 0.0010-0.001, I2 = 0.0%, P = 0.847). A total of 2 patients required re-operation, including one for device dislocation and the other for wound infection. Device dislocation may be a procedure-related complication caused by a lack of experience. No cases of new mild or significant aortic regurgitation were observed. The pooled rates of mild-to-significant aortic/tricuspid regurgitation and severe arrhythmias were 0.000 (95% CI: 0.000–0.000, I2 = 0.00%, P = 1.0). The pooled rate of severe complications in the follow-up period was 0.000 (95% CI: − 0.000-0.000, I2 = 0.0%, P = 1.00). cAVB is a severe complication during and after device closure for a VSD, especially in cases with perimembranous VSD (pmVSD) [19, 20]. However, no cAVB occurred in the enrolled studies, and the pooled rate was 0.00 (0.00–0.00) (95% CI: − 0.000-0.000, I2 = 0.0%, P = 1.0). We attributed this to the relatively long distance between the conduction tissue and the rim of the dcsVSD. The posteroinferior margin of the dcsVSD is usually well separated from the tricuspid valve annulus by a band of muscle [22]. Thus, the chance of conduction system injury from mechanical trauma compression by the delivery system or device seems to be small [23]. However, we should still pay attention to late cAVBs resulting from chronic inflammation or fibrosis.

Minor complications include trivial to mild residual shunts (RSs), AR, pulmonary regurgitation (PR). Most studies couldn’t provide information regarding intraoperative minor complications. Thus, we would discuss the minor postoperative complications and its outcome in follow-up period.

The pooled rate of postoperative residual shunts was 0.050 (95% CI: 0.050–0.085; I² = 74.5%, P = 0.000). However, most of them disappeared during the follow-up period, and the pooled rate of follow-up RSs was 0.00 (95% CI: − 0.001-0.001; I² = 0.00%, P = 0.51). This change means that most RSs could close spontaneously during the follow-up period. We attribute this to endothelialization covering the surface of the device and neointima forming several weeks after operation [24]. Further meta-regression showed that a ratio of occluder size (mm)/weight (kg) above 0.4 was a risk factor for postoperative RSs with a Coef of − 0.28 (95% CI: − 0.49- -0.66, p = 0.020). Zhang and his colleagues also reported that procedure-induced aortic regurgitation (AR) was associated with device diameter and patient weight (OR = 12.3 95% CI 1.5–99.2) [16]. Thus, the ratio of occluder size (mm)/weight (kg) below may be another patient selection criterion for perventricular device closure of dcsVSDs.

The pooled rate of postoperative AR was 0.045 (95% CI: 0.018–0.071, I² = 50.96%, P = 0.00). Further meta-regression showed that a ratio of VSD size (mm)/weight (kg) above 0.3 was a risk factor for postoperative AR, with a Coef of − 0.14 (95% CI: − 0.20-0.87, P = 0.001), which suggested that patients with a low body weight and a relatively large VSD size are more likely to experience postoperative AR. Most of the AR disappeared during the follow-up period, and the pooled follow-up rate of AR was 0.001 (95% CI: − 0.003-0.004, I² = 63%, P = 0.009). However, new AR also occurred during the follow-up period. The position of the aortic cusp changes throughout the cardiac cycle. During the diastolic phase, the mid portion of the cusps will be pushed below the annulus level by the diastolic blood pressure. Although the device has a 0-mm superior rim, the interface between the device and the aortic cusps may be unavoidable after endothelialization during the follow-up period.

The pooled rate of trivial to mild postoperative pulmonary regurgitation (PR) was 0.002 (95% CI: − 0.004-0.008, I² = 57.10%, p = 0.030). However, the pooled rate of trivial to mild pulmonary regurgitation was 0.003 (95% CI: − 0.004-0.009, I² = 63.30%, p = 0.012). The pooled rates of trivial to mild postoperative/follow-up pulmonary regurgitation were similar. Although the pulmonary system has a relatively low pressure, trivial to mild PR is acceptable. However, a longer follow-up may be necessary to assess the true impact on pulmonary valve function. Further student t-regression showed that a ratio of occluder size (mm)/weight (kg) above 0.4 was a risk factor for follow-up PR with a Coef of − 0.28 (95% CI: − 0.49- -0.66, P = 0.020), which suggested that a ratio of occluder size (mm)/weight (kg) below 0.4 may be another patient selection criterion for perventricular device closure for dcsVSDs.

Perventricular device closure may be an alternative to conventional surgical repair in selected patients with dcsVSDs. The success rate wasn’t related to publication time and sample size, suggesting a relatively short learning curve and the technique’s potential for application. The incidence of severe complications during the hospital stay and follow-up was low, especially for severe valvular regurgitation, device dislocation and cAVB. However, the indications and contraindications are still doubtful. Patients with a VSD size of 5 mm^− 10 m may be suitable for this procedure. Additionally, a ratio of occluder size (mm)/weight (kg) above 0.4 may be another contraindication. Further studies should include a larger number of cases with longer follow-up periods and sufficient data to determine the risk factors for procedure failure.