Surgical Closure of Multiple Muscular Ventricular Septal Defects in Children Using 3D-Printed Models

Introduction

Multiple muscular ventricular septal defects (VSDs) are defined as three or more defects in any one portion of the muscular interventricular septum (IVS). ¹ Serraf et al define “Swiss cheese” septum as having four or more defects. ² These defects are often located in parts of the IVS which are difficult to access surgically. Identifying all the defects intraoperatively and accurately delineating their margins remains the main surgical challenge, without which residual defects are inevitable. When viewed from the right ventricle (RV), the trabeculations of the RV obscure the actual margins of the defects. When viewed from the left ventricle (LV), they appear as discrete, often single defects with easily identifiable margins. ³ This raises the possibility of a single defect on the LV side of the IVS appearing as multiple defects on the RV side of the IVS. ⁴ The surgical management of multiple VSDs is complicated by operative mortality, heart block, arrhythmias, ventricular failure, and residual defects requiring reintervention.^1–3 Various techniques have been described to identify and close these defects with acceptable results, but no one technique has shown superiority over others.^5–17 The choice of technique used to locate and close these defects determines the surgical outcome. ⁴

Three-dimensional (3D)-printed models are widely used in various surgical fields, notably orthopedic and maxillofacial surgery.^18,19 More recently, they have also proven helpful in managing complex congenital heart defects due to their ability to demonstrate intracardiac anatomy and scale accurately. Double outlet RV has been the main indication for using 3D models, but they have also been used for other conditions.^20–27 We have used this technology for the surgical management of multiple muscular VSDs and present our initial clinical experience.

Patients and Methods

Between September 2021 and July 2023, all patients with multiple muscular VSDs for whom a 3D-printed model was used to aid surgical planning were included in the study. Perioperative data were collected retrospectively. Along with routine preoperative investigations, which included an echocardiogram, all patients underwent a preoperative cardiac computerized tomography (CT) scan from which the cardiac radiologist compiled 3D images.

The details of the CT scan technique have been previously described. ²⁸ Each CT scan was performed under general anesthesia on a Siemens SOMATOM Force scanner (Dual Source 384 [2 × 192] slice, Siemens) with prospective electrocardiogram (ECG) gating triggered in diastole; 2 mL/kg of contrast (Omnipaque 300, GE Healthcare Inc.) was administered with the rate of infusion individualized to the child's weight and heart rate. Raw data sets were then individually worked-up on a Siemens Syngo application MM reading and AGFA enterprise Patient Archiving and Communication Systems (The Agfa-Gavaert Group). Two-dimensional and 3D reformats of all relevant structures were performed on both systems. The virtual 3D model was then cropped to the blood pool, and the myocardium was nulled.

The raw CT data were then sent to the Australian Centre for Complex Integrated Surgical Solutions team at the Translational Research Institute for 3D modeling in Brisbane. The details of the technique used to create the 3D models have been previously published. ²⁸ The CT scan was imported into Materialise Mimics, a 3D medical image-based engineering software (Medical v25.0, Materialise). Using a single threshold minimum of 223HU, the blood pool was segmented from the surrounding tissues. After identifying the VSDs in the 3D render, the blood pool regions on the outer ventricle walls were filled, and areas outside the main cardiac volume were cropped and removed from the model. The resultant blood pool was then converted to a 3D solid and exported to Materialise 3Matic (Medical v17.0, Materialise). A hollow model was generated with the internal space defined by the blood volume.

A datum plane was aligned to the septum, as evidenced by the blood pool, and the 3D model was then split into three segments aligned to the ventricular septum. The datum plane was then offset to the left and right to a distance such that the entire septal wall could be visualized from either side for 10 to 15 mm. These outer two planes were then used to divide the 3D model into three segments (left, septal, and right). The virtual 3D model was used to precisely define the anatomy of the VSDs (Figure 1). The models were then 3D printed in white polylactic acid on a FlashForge Pro 2 3D printer (Zhejiang Flashforge 3D Technology Co., Ltd). The printed model was split into three segments with a plane of cleavage positioned on either side of the septum, thereby exposing the anatomy of the defects in the IVS. This preserves the spatial relationships between the VSDs, adjacent muscle bundles, and the surface of the heart (Figure 2). All 3D data sets were individualized to the patient and reviewed jointly at every step by the cardiac radiologist, cardiac surgeon, and 3D engineer. The model was then used to identify the multiple VSDs and plan surgical intervention precisely.

OPEN IN VIEWER

Figure 1.

The cardiac radiologist compiled preoperative cardiac computerized tomography (CT) scan images and imported them into a specialized software package (Materialise Mimics^®, medical v25.0, Materialise). The software was used to create 3D images accurately identifying the location of the muscular ventricular septal defects (VSDs) (arrows). Using specialized software, the 3D engineer used this data to construct a patient-specific 3D heart model.

OPEN IN VIEWER

Figure 2.

The central section of the model incorporating the interventricular septum is used to accurately identify the location of the multiple muscular ventricular septal defects and the anatomy of the adjacent muscle bundles. There is often a single confluent defect, split into multiple defects by crossing muscle bundles on the right side of the septum when viewed from the left side of the septum.

Results

Six patients underwent closure of multiple VSDs during the study period using a 3D model. Table 1 gives the baseline characteristics for each patient. The mean age at surgery was 3.5 years (SD ±2.8 years). The mean weight was 13.3 kg (SD ±24 kg). Five (5/6, 83.3%) patients had previously undergone PA banding at a mean age of 63.2 days (SD ±75.8 days). Three patients (3/6, 50%) had a previous aortic arch repair. The VSDs were approached through the RA in three (3/6, 50%) and the RV in three (3/6, 50%) patients. Additional procedures included PA debanding (n = 5) and aortic valve repair (n = 1). Mean cardiopulmonary bypass and myocardial ischemia times were 185.2 min (SD ± 94.8 min) and 147.5 min (SD ± 86 min), respectively. Table 2 gives the perioperative and postoperative data for each patient. Postoperatively, no patient developed heart block or had a hemodynamically significant residual VSD. There were no postoperative complications in four patients. One patient developed necrotizing enterocolitis, which was managed conservatively, but had to be readmitted after being discharged from the hospital for intestinal obstruction requiring bowel resection. One patient, who had supraventricular arrhythmia preoperatively requiring transvenous pacing, developed the same arrhythmia postoperatively, which was managed medically. There were no hospital deaths.

OPEN IN VIEWER

Table 1.

Baseline Characteristics, Preoperative Data.

	Patient 1	Patient 2	Patient 3	Patient 4	Patient 5	Patient 6
Age at surgery (years)	6.9	6.6	1.3	3.9	0.3	2.1
Sex	Male	Male	Male	Male	Female	Female
Weight (kg)	17.7	17.9	14	15.5	3.4	11.4
Diagnosis	Multiple muscular VSDs, PDA	Multiple muscular VSDs, coarctation of aorta, hypoplastic transverse arch, secundum ASD, moderate aortic stenosis, mild aortic incompetence	Multiple muscular VSDs, secundum ASD, hypoplastic arch with juxtaductal coarctation	Mid muscular VSD, small ostium secundum ASD	Multiple muscular VSDs, secundum ASD, atrial ectopic tachycardia and bradycardia	Muscular VSD, interrupted aortic arch (type A)
Previous procedure(s)	PA banding, ligation of PDA	Repair of hypoplastic aortic arch, PA banding	Repair of hypoplastic arch and, PA banding	PA banding	Transvenous pacing	Repair of interrupted aortic arch, PA banding
Non cardiac abnormalities	Mosaic trisomy 16, ex premature 34 weeks gestational age	None	None	None	Renal dysfunction, mechanical ventilation, hepatic dysfunction, rhinovirus infection, ventilator associated pneumonia	None
Location of VSDs	Mid muscular, anterior, and apical septum	Anterior and apical septum	Anterior and apical septum	Mid to lower muscular septum	Posterior muscular septum	Mid muscular septum

Abbreviations: ASD, atrial septal defect; PA, pulmonary artery; PDA, patent ductus arteriosus; VSD, ventricular septal defect

OPEN IN VIEWER

Table 2.

Perioperative and Postoperative Data.

	Patient 1	Patient 2	Patient 3	Patient 4	Patient 5	Patient 6
VSD approach	RV	RV	RV	RA	RA	RA
Procedure	VSD closure, PA debanding, PA reconstruction	VSD closure, ASD closure, PA debanding, PA reconstruction, Aortic valve repair, reduction aortoplasty (ascending aorta)	VSD closure, ASD closure, PA debanding, PA reconstruction	VSD closure, ASD closure, PA debanding, PA reconstruction	VSD closure, fenestrated ASD closure, PDA ligation	VSD closure, ASD closure, PA debanding, PA reconstruction
Post op ventricular function	Normal RV/LV function	Normal RV/LV function	Normal RV/LV function	Normal RV/LV function	Normal RV/LV function	Normal RV/LV function
Hemodynamically significant residual defects	No	No	No	No	No	No
Postoperative heart block	No	No	No	No	No	No
Follow up echocardiogram	Normal LV/RV function, no residual VSD	Normal LV/RV function, no residual VSD, Moderate AS, Moderate AR	Normal LV/RV function, no residual VSD	Normal LV/RV function, hemodynamically insignificant apical VSD	Normal LV function, hemodynamically insignificant apical and anterior muscular VSD	Normal LV/RV function, no residual VSD

Abbreviations: AR, aortic regurgitation; AS, aortic stenosis; ASD, atrial septal defect; LV, left ventricle; PA, pulmonary artery; PDA, patent ductus arteriosus; RA, right atrium; RV, right ventricle; VSD, ventricular septal defect;

All six patients had a normal biventricular function at discharge and at a median follow-up duration of 1.7 months (IQR 1.2-7.4 months), and none had hemodynamically significant residual VSDs.

Comments

Complete closure of all multiple muscular VSDs with low operative risk while avoiding heart block, arrhythmias, and ventricular dysfunction should be the goal of the surgical management of these defects. ¹⁶ The myriad techniques to identify and close these defects reflect the challenge of achieving these goals with a single technique. Significant residual defects contribute to both early and late morbidity and mortality. ² Smaller defects are often masked by the magnitude of shunting across the larger defects, and hence, accurately locating all the defects preoperatively or by intraoperative real-time echocardiogram can be challenging.^3,16 Alsoufi et al recommended intraoperative exploration to detect all defects, as echocardiography can often underestimate the number of defects. ¹⁶ Exploration can be done from the right side of the septum or the left (across a perimembranous defect or the atrial septum) by gently probing the septum or by the illumination method.^3,9,16,29 Localization with a guide wire placed intraoperatively or a catheter placed through the defect preoperatively has also been described, along with pressurizing the LV with blood.^15,30,31 While searching for the defects, extensive division of muscle bundles can compromise the RV and septal function, and multiple, incorrectly placed incisions can cause significant impairment of RV function. This emphasizes the importance of accurately localizing these defects preoperatively. ¹⁵

Depending on the location of the defects, the VSDs can be approached through the RA, RV, or LV. While the approach through the RA can be considered safe compared with a ventriculotomy, it is not always possible to identify all defects through the RA, especially apical defects, and the more anterior defects, which are better visualized through a ventriculotomy.^2,4,16 However, a ventriculotomy is not without the risk of morbidity and mortality, as described by Serraff et al and Alsoufi et al, further underscoring the importance of accurate localization of the VSDs preoperatively to minimize the extent of the ventriculotomy.^2,16

Because each 3D model is patient-specific, it precisely replicates the intracardiac anatomy, including the anatomy of the muscle bundles crossing the defect and the appearance of the defect when viewed from both sides of the IVS. Its geometric and spatial precision allows the surgeon to thoroughly study the intracardiac anatomy and plan the surgical approach. ²⁴ Compared with the other forms of 3D imaging of the heart (echocardiogram, CT scan), the physical 3D model also allows the surgeon to physically examine the model, enhancing the surgeon's knowledge and understanding of the anatomy of the defects. The model can also be split on either side of the IVS to give a clear view of the VSDs both from the right and left sides of the IVS (Figure 2). If a ventriculotomy is planned, the precise location of the incision directly above the VSDs can be planned, thus limiting the size of the incision and avoiding incorrectly placed incisions. The predictability of the 3D model also reduces the time spent “searching” for the defects when the heart is arrested. Appreciating the anatomy of the muscle bundles crossing the defects is another benefit of using a 3D model that accurately replicates the intracardiac anatomy. Division of these muscle bundles often converts the multiple defects, when viewed from the right side of the septum, into a single confluent defect, which then aids in the accurate closure of the VSD with a patch. None of the patients in our series had any significant residual defects on postoperative echocardiography, and the short ventriculotomy did not appear to have caused significant right ventricular dysfunction (albeit with a short duration of follow-up).

Although CT exposes the patient to radiation, we prefer to use CT scans because they have better spatial resolution than magnetic resonance imaging (MRI), especially for small children with multiple septal defects. An MRI also requires a longer period of general anesthesia. We also recommend using a printed 3D model. We attempted to use only a computer-generated 3D image in one patient but reverted to using printed models because the tactile element in the decision-making process cannot be discounted, and the physical model is easier to use in the operating theater. Nevertheless, with experience and based on user preference, it may be possible to transition to only a computer-generated 3D image and achieve the same surgical result.

The 3D model undoubtedly assists in deciding which VSDs can be closed or should be closed. With experience, we have further modified the 3D model, making it even more useful for discerning which VSDs merit closure by separating any mass of overlying muscle bundles as an individual component from the septum. This facilitates a decision about whether any residual VSDs are likely to be rendered hemodynamically insignificant by dense overlying muscle bundles.

Occasionally, deciding to proceed with a primary biventricular repair with multiple muscular VSDs can be difficult, prompting palliation with a PA band to allow for growth and reassessment later. Hypertrophied RV muscle bundles following placement of a PA band can cause difficulty in the exposure of the defects and postoperative morbidity. ⁹ Single-stage repair of multiple muscular VSDs has been previously described.^3,16,31 After carefully reviewing all available information from the preoperative imaging, the surgeon must decide whether all the defects can be closed to facilitate single-stage repair. Using a 3D-printed model can further aid the decision-making process in both instances.

Conclusion

Surgical closure of multiple muscular VSDs continues to be challenging, and the congenital cardiac surgeon should be aware of the various techniques available to manage these defects. Patient-specific 3D-printed models can aid decision-making and surgical planning. In our experience, using 3D printing to assist the closure of multiple muscular VSDs is safe, reliable, and reproducible. We recommend adding 3D printing to the surgeons’ armamentarium when dealing with the challenge of closing multiple muscular VSDs in children.