The role of three-dimensional printed cardiac models in the management of complex congenital heart diseases

Abstract

Background

Three-dimensional printing is a process enabling computer-assisted conversion of imaging data from patients into physical “printed” replicas. This has been extrapolated to reconstructing patient-specific cardiac models in congenital heart diseases. The aim of this study was to analyze the impact of three-dimensional printing in surgical decision making in selected cases of complex congenital heart disease by creating patient-specific printed models.

Methods

Patients with complex congenital heart diseases with unresolved management decisions after evaluation by echocardiography, cardiac catheterization, and cardiac computed tomography were included with intent to aid in surgical decision making. Three-dimensional models were created from computed tomographic images by an outsourced firm using computer applications. All cases were reviewed by the same team before and after the cardiac models were prepared. The management decisions were grouped as either “corrective surgery” or “no surgery or palliation” The impact of the surgical decision pre and post three-dimensional cardiac model was analyzed by applying Cohen’s kappa test of agreement.

Results

Ten patients were included, of which five were of increased pulmonary blood flow, and five were of decreased pulmonary flow. The commonest indication for three-dimensional printed models was to establish the routability of the aorta and pulmonary artery to their respective ventricles (in five patients). The nonagreement between the decision taken before and after the cardiac model was 80%, with kappa −0.37 and P value 0.98.

Conclusions

Three-dimensional printed cardiac models contribute to better decision making in complex congenital heart diseases enabling safer execution of any complex congenital heart surgery.

Introduction

The prevalence of congenital heart diseases (CHD) is about 10–12 per 1000 live births.¹ CHD have a wide anatomical spectrum. They may be single defects like ventricular septal defects (VSD), atrial septal defects (ASD), or complex anomalies like double outlet right ventricle (DORV), congenitally corrected transposition of great arteries (CCTGA). A small subset of CHD are extremely complex. They present in a myriad combination of structural cardiac anomalies, as well as abnormalities of spatial relationship to the thoracic cavity. The heart is a dynamic and multidimensional organ, and the available means of evaluation of the cardiac anatomy are echocardiography, computed tomographic (CT) angiography, cardiac catheterization, and cardiac magnetic resonance imaging (MRI). These modalities of imaging are sufficient in most of the cases for surgical decision making. However, a few complex congenital heart diseases have an anatomy that cannot be accurately assessed by the two-dimensional images of echocardiography, CT scan, and MRI for surgical decision making. This translates to an unchartered territory for the surgical team before embarking on surgery under cardiac bypass (CPB). The best surgical outcomes can only be achieved by an accurate preoperative understanding of the complex three-dimensional (3D) spatial relationship. An accurate replica of the patient’s abnormal heart is an ideal method of visualizing the anomaly and approach to the CHD. This is made possible by three-dimensional (3D) printing. 3D printing or additive manufacturing is a process of conversion of digital imaging data into hard replicas by adding layers of material that are fused additively together. It was first described in the early 1980s by Charles Hull, who called it “stereolithography.” This technology has been used for several industrial applications and also in other branches of medicine like orthopedics and dentistry.² We carried out this study to evaluate the role of 3D printing of cardiac models on surgical decision making in complex CHD at a tertiary level high volume pediatric cardiac center of the Indian Armed Forces.

Aims and objectives

Aim

To analyze the impact of 3D printing of patient-specific cardiac models on the decision making in the management of complex CHDs.

Objectives

• •To compare the management decision made by conventional imaging data vs decision made by 3D printed cardiac model in the same patient.
• •To identify the common subsets of complex congenital heart diseases that require 3D printed cardiac models for decision making in surgical management.

Materials and methods

The study was conducted over 2 years at a tertiary level high volume pediatric cardiac center. Patients of complex CHDs with unresolved surgical decisions by echocardiography, cardiac catheterization, and CT angiography were included. The selection of patients to undertake a 3D model was taken by the pediatric cardiac care team comprising the pediatric cardiologist and pediatric cardiothoracic surgeon. A written informed consent was obtained from all the parents. All patients underwent standard cardiac imaging by two-dimensional echocardiography sweep views and color Doppler interrogation^3,4 on Philips IE 33 (Philips Healthcare, Netherlands) echocardiography machine. The patients underwent cardiac catheterization by a pediatric cardiologist where indicated.⁵ They underwent cardiac computed tomography under short sedation by an anesthesiologist where required.⁶ CT scan was done on 256 slice scanner (Philips Inc, The Netherlands. Brilliance ICT 256) with bolus tracking with 1 mL/kg of iodinated nonionic contrast material intravenously. The slice thickness of 0.9 mm with a slice increment of 0.45 mm was applied. The average dose of radiation per study was 498 mGy. The 3D printing for CT scan images was outsourced to M/s Global Healthcare, New Delhi. The computer application used was Materialise Mimics, Materialise NV, Leuven, Belgium. This computer application is used to create a wall around the blood volume to allow visualization of intracardiac structures. The data was exported in the “standard tessellation language” (STL) format. 3D models were printed using a 3D printer with Fusion Deposit Modelling (FDM) or selective laser sintering (SLS) technique (polyamide). The material used for prints was Poly Lactic Acid (PLA) or Poly Vinyl Alcohol (PVA). The models were prepared with paraseptal cuts to study the endocardial structures and interiors of the chambers. The patients were re-evaluated by the same team along with the physical inspection of the 3D printed heart model (Fig. 1, Fig. 2). The final decision of management was taken based on the model. The decision of management plan was grouped as two outcome variables; one variable was “corrective surgery” and the other “palliative surgery or no surgery”. The decisions of management before and after 3D cardiac model was compared. Statistical analysis was done on SPSS Statistics Version 24 by applying Cohen’s kappa test of agreement. The percentage nonagreement between both the decisions was taken as a measure of the impact of 3D cardiac model on the decision making. The kappa value > 0 and P value < 0.05 was taken as statistically significant.

Fig. 1.

The comparison of the 3D reconstructed image of CT scan and 3D model of the heart of a patient with DORV, VSD, PS. 1A. The anterior view of the 3D reconstructed image of CT scan where the aorta (Ao) appears routable. 1B. The lateral view of the 3D reconstructed image of CT scan where the aorta appears doubtfully routable. 1C. The posterior view of the 3D reconstructed image of CT scan where the aorta appears nonroutable. 2A. Anterior view of the model in anatomical position and aorta arising from the anterior (Ao) right ventricle (RV). 2B. Lateral view of model assembly held by magnets showing the available paraseptal cuts for dismantling the model and visualizing internal structures. 2C. The model with the anterior-most segment removed showing the interventricular septum (IVS), VSD (arrow) and the opening of the aorta (Ao). The aorta (white arrowhead) is completely committed to the RV, therefore being nonroutable. 2D. The model with the anterior two segments removed and the left ventricle (LV) displayed surface showing VSD (arrow).

Fig. 2.

The comparison of the 3D reconstructed image of CT scan and 3D model of heart of a patient with congenitally corrected transposition of the great arteries with VSD (Post PA Band). 1A. The anterior view of the 3D reconstructed image of the CT scan where it appears correctable. 1B. The posterior view of the 3D reconstructed image of the CT scan where it appears correctable. 1C. The lateral view of the 3D reconstructed image of CT scan where it appears correctable. 1D. The lateral view of the 3D reconstructed image of the CT scan where it appears correctable. 2A. The anterior, posterior, and lateral view of the model in anatomical position. 2B. The model with the left ventricular free wall segment removed showing the VSD (arrow) and the opening of the aorta (Ao) and pulmonary artery (post PA band) (PA). 2C. The model with the left ventricular free wall segment removed showing the VSD and the opening of the aorta (Ao) and pulmonary artery (post PA band) (PA). The relationship (white arrowhead) of the aorta, VSD and PA shows that the lesion is not correctable.

Results

Ten patients were included in the study over two years, of which six were male and four female. The average age of the patients was 38.2 months (±40.6 months). Five patients were of increased pulmonary blood flow, and five were of decreased pulmonary blood flow physiology. Six patients underwent cardiac catheterization; the indication for the same was obtaining hemodynamic information for the assessment of pulmonary vascular resistance in one (patient number 7) and for suitability for a single ventricle pathway in four. One patient with Scimitar syndrome underwent cardiac catheterization for coil embolization of abnormal aortopulmonary collateral vessel. The commonest indication for 3D printed models was to establish the routability (in five patients) of the great vessels, namely the aorta and pulmonary artery, to their respective ventricles (Table 1). The other indications were to assess the pulmonary venous anatomy in two patients, pulmonary arterial anatomy in one, assessment of MAPCA anatomy in one, and understand the overall complex cardiac anatomy in another patient.

Table 1.

The clinical characteristics of the study population, underlying diagnosis, indication for 3D cardiac model, pre and post 3D model management decision, and the change in management decision after 3D model.

Patient.No	Age (Months)	Pulmonary arterial flow physiology (Low flow vs high flow)	Diagnosis on Echocardiography	Indication for 3D model	Pre 3D model surgical decision	Anatomical component in 3D model in variation to echocardiography data	Re assessment of decision based on 3D model	Surgical procedure carried out	Change in decision after Model (Yes/No)
1.	24	Low pulmonary arterial flow	DORV,VSD,PS - Post RMBTS	Routability of great vesselsa to their respective ventricles	“Corrective surgery”- Routable great vesselsa to their respective ventricles	Ventriculoarterial relationship	“ Palliative surgery”- Not routable great vesselsa to their respective ventricles	Bi-directional Glenn	Yes
2.	60	High pulmonary arterial flow	PAPVC,ASD,PAH	Anatomy of pulmonary vein	“Palliative surgery”-Right upper pulmonary vein not routable directly and needs reimplantation	Pulmonary venous drainage	“Corrective surgery”- Right upper pulmonary vein routable directly	Rerouting of PV,ASD Closure done	Yes
3.	49	Low pulmonary arterial flow	TGA,VSD,PS-Post RMBTS	Routability of great vesselsa to their respective ventricles	“Palliative surgery”-Not routable great vesselsa to their respective ventricles	Ventriculoarterial relationship	“Corrective surgery”- Routable great vesselsa to their respective ventricles	Nikaidoh procedure	Yes
4.	18	Low pulmonary arterial flow	DORV,VSD,PAH, post PA band	Routability of great vesselsa to their respective ventricles	“Corrective surgery”-Routable great vesselsa to their respective ventricles	Ventriculoarterial relationship	“No surgery”-Not routable great vesselsa to their respective ventricles	Medical management	Yes
5.	9	Low pulmonary arterial flow	DORV,VSD,PS	Routability of great vesselsa to their respective ventricles	“Corrective surgery”-Routable great vesselsa to their respective ventricles	Ventriculoarterial relationship	“Palliative surgery”-Not routable great vesselsa to their respective ventricles	Bi-directional Glenn	Yes
6.	140	High pulmonary arterial flow	TOF (LPA from Ascending Aorta)	Pulmonary artery anatomy	“Corrective surgery”- Favourable anatomy for corrective surgery	Pulmonary artery anatomy	“No surgery”-Inoperable due to hemodynamic factors	Medical management	No
7.	9	High pulmonary arterial flow	DORV, VSD, Double Aortic Arch. Pulmonary arteries from right arch	Complexity of Anatomy	“No surgery”-Not correctable anatomically	Pulmonary artery anatomy	“No surgery”-Not correctable	Medical management	No
8.	14	High pulmonary arterial flow	Scimitar syndrome	Pulmonary venous anatomy	“Corrective surgery”-Routable anomalous pulmonary veins to left atrium	Pulmonary venous drainage	“No surgery”-Inoperable	Coil embolization of MAPCA	Yes
9.	9	Low pulmonary arterial flow	Pulmonary atresia,VSD,MAPCA’S	Aortopulmonary Collateral arterial assessment	“Corrective surgery”-Operable	Pulmonary artery anatomy	“No surgery”-Inoperable	Medical management	Yes
10.	50	High pulmonary arterial flow	CCTGA, VSD Post PA Band	Routability of great vesselsa to their respective ventricles	“Corrective surgery”-Routable great vesselsa to their respective ventricles	Ventriculoarterial relationship	“No surgery”-Unfavourable anatomy for surgery- The relationship of the aorta, VSD and pulmonary artery was not suitable for corrective surgery	Medical management	Yes

Abbreviations: 3D -Three dimensional, CHD - Congenital heart disease, CT - Computed tomography, MRI - Magnetic resonance imaging, CPB - Cardiopulmonary Bypass, ACC - Aortic Cross Clamp, DORV - Double outlet right ventricle, VSD - Ventricular septal defect, ASD - Atrial septal defect, PS - Pulmonary stenosis, PAH - Pulmonary artery hypertension, TOF - Tetralogy of Fallot, RMBTS - Right modified Blalock Taussig shunt, MAPCA - Major aortopulmonary collateral arteries, LPA - Left pulmonary artery, RPA - Right pulmonary artery, PAPVC - Partially anomalous pulmonary venous return, CCTGA - Congenitally corrected transposition of great arteries, PA Band - Pulmonary artery band.

^aRoutability of great vessels to their respective ventricles is and important decision that differentiates between complete correct ability vs palliative single ventricle pathway surgeries.

Correlation of the cardiac anatomy with the 3D printed model

It was observed that the 3D prints had a good correlation with surgical anatomy on the operating table in the five patients who underwent surgical procedures.

Impact on decision making

The decision of management changed after the 3D model from the pre 3D model decision in eight patients out of ten. The nonagreement between the decision before and after 3D model was 80% (agreement of 20%). The kappa value - 0.37 (<0) and the P value is 0.98 (>0.05).

The details of diagnosis, indications for 3D model, outcome variables, the final decision of management, and the surgical procedures carried in the patients are enumerated in Table 1. All patients had optimal postoperative recovery with an uncomplicated course and zero mortality upto their six months follow up.

Discussion

The traditional methods of structural evaluation in CHD are echocardiography and cardiac catheterization.^{3, 4, 5} Lately, due to the need for the precise imaging of cardiac anatomy, CT scan and MRI have been increasingly employed.⁷ In addition, advances in surgical techniques have enabled surgical treatment of several complex CHD, thereby increasing the need for greater precision in surgical planning.⁸ The need for precise delineation of cardiac anatomy is essential for surgical decision making prior to surgery under CPB. The time under CPB directly influences postoperative recovery of the myocardium.

Our center has an average congenital cardiac surgical load of about 300 per year. Echocardiography, along with CT scan or MRI, is sufficiently informative in most of the CHDs but falls short in a small subset of complex CHDs. Over two years, ten patients with complex CHDs were taken up for 3D model reconstruction with an intent to aid pre surgery decision making. This is one of the largest single-center study carried out to assess the impact of 3D cardiac model in presurgery decision making in complex CHD.⁹ Valverde et al reported that 3D models in complex CHDs changed the surgical decision in 19 cases out of 40 in a multicentric study. (47.5% of the cases, confidence interval of 29.6–61.5%).¹⁰ We noted that in eight out of ten patients, the decision changed after evaluation of the 3D. There have been no studies carried out to evaluate 3D cardiac model as a diagnostic tool. Therefore unlike other workers, we applied the tests of the agreement for the decision taken before and after the 3D model, which reflected the finality of the diagnosis. The disagreement or difference in the decisions pre and post 3D model was noted to be 80% making it clinically significant as a diagnostic tool. But, the Cohen’s kappa value noted is below 0 and p > 0.05, thereby making it statistically not significant. The most probable reason for this is the small sample size. The nonhomogeneity of CHDs and the number of such complex CHDs being considered for surgical correction are limited; hence, the restrictions of small sample sizes are difficult to overcome. Rarely the sample size has exceeded ten in most of the available literature (Table 2).⁹ Nevertheless, 3D cardiac models have an increasing role in the surgical planning of CHD in the coming future. Therefore we undertook this study to evaluate it as a diagnostic tool for the first time. Since it is a limited resource, we also attempted to identify the likely CHD that may require 3D model for decision making. We analyzed the subsets of CHDs for which 3D models have been utilized by other workers. DORV with VSD is one subset that has been a center of debate for surgical planning. 3D models have been used for decision making and training alike for this lesion by several workers.^{11, 12, 13} 3D models aid in visualizing the VSD and its relationship to the great arteries to decide the routability of great vessels to the respective ventricles. In our study, the commonest indication for undertaking 3D model reconstruction was also to establish routability. We had three cases of DORV with remote VSD, one case of CCTGA, VSD and one case of TGA with remote VSD with pulmonary stenosis. DORV has been exclusively studied by Garekar et al from India. Garekar et al utilized 64-detector scanner (Philips Healthcare) CT scan without elctrocardiography (ECG) gating in some cases, and MRI in some cases for acquiring images¹³ (Table 1). Whereas, we employed a CT scan only on a 256 slice detector with ECG gating to acquire images. The CT scan could be done rapidly with a short sedation without the requirement of breath holding viz a viz MRI. The images obtained could reconstruct all the chambers, vascular anatomy, and endocardial structures in fine detail. The advantage MRI has to offer over CT scan is a better myocardial study and providing hemodynamic data interpretation. However, for a study of spatial orientation of cardiac structures, MRI offers no additional advantage.

Table 2.

A comparison of similar studies conducted on 3D printed cardiac models in congenital heart diseases.

Author	Year	Multicentric/single center	Sample size	Criteria of inclusion	Modality of imaging for 3D reconstruction	Subtype of congenital heart disease studied
Garekar et al.	2016	Single center	6	Consecutive patients	Cardiac MRI and CT scan	DORV with VSD
Valverde et al.	2017	Multicentric – 10 centers	40	Case crossover of patients in whom the surgical decision was not reachable by conventional imaging	Cardiac MRI and CT scan	Multiple CHD, predominantly DORV with VSD
Kappanayil M et al.	2017	Single center	5	Case crossover of patients in whom the surgical decision was not reachable by conventional imaging	Cardiac MRI and CT scan	DORV with VSD
Bhatla P	2017	Single center	6	Consecutive patients	Cardiac MRI and CT scan	DORV with VSD
Xu J et al.	2019	Single center	17	Consecutive patients	CT scan	Abnormal pulmonary venous drainage
Tiwari N et al. (Present study)	2020	Single center	10	Case crossover of patients in whom the surgical decision was not reachable by conventional imaging	CT scan (256 slice)	DORV with VSD and other CHD s of ventriculoarterial discordance

The other subset of CHDs included in our study that pose a challenge for imaging, as well as surgery, was abnormal pulmonary venous drainage. We had two patients with abnormal pulmonary venous drainage. Xu et al studied seventeen 3D reconstructed models that contributed to improving surgical planning for anomalous pulmonary venous drainage.¹⁴ Pulmonary artery and its branches are the other vascular structures that need additional information. In our study, we had one patient with pulmonary atresia with VSD (Fig. 3), and one patient had a congenital unilateral absence of branch pulmonary artery with MAPCAs supplying the affected lung. The feasibility for complete correction was studied by 3D model reconstruction. Another patient had complex anatomy with DORV, large VSD, double aortic arch with branch pulmonary arteries arising from the right-sided arch. This complex anatomy was the most difficult to reconstruct among the ten cases (patient No 7 in Table 1).

Fig. 3.

3D printed cardiac model of a patient with pulmonary atresia with no native branch pulmonary arteries. The ascending aorta (In rectangle) and the arch with the descending aorta (circle) can be separated for better visualization. The lungs are supplied by the major aortopulmonary collateral arteries arising from the descending aorta (Arrows).

A comparison of similar studies of the role of 3D printed cardiac models in congenital heart diseases is tabulated in Table 2. It is worthwhile to note that our study is a large single centric study that has studied a variety of CHDs. Our study demonstrates that using only a CT scan can provide informative 3D models of a variety of CHDs. In addition, being a single-center study, the observer bias is limited because only one team was involved in the study.

The presence of a 3D model in preoperative surgical planning allowed the team to make a clear strategy in these cases. The comparison of the 3D reconstructed images of the CT scan, and the actual 3D models demonstrated the advantage of a 3D model (Fig. 1, Fig. 2). The surgeon did not need to “rethink and improvise on the table,” thus saving operating time on the arrested heart. Valverde et al have statistically demonstrated the advantage and accuracy of 3D models as an extension to cardiac imaging in the management of complex CHD.¹⁰ However, no controlled trial could be conducted to conclusively establish its benefit by any worker. The patients with complex CHDs who pose a dilemma for management are very uncommon, and those who underwent a complete correction in our study were unique. Hence no similar cases were available for comparison of surgical outcomes. However, we could clinically attribute the faster postoperative recovery to the right decision and accurate preoperative planning. The cost of 3D printing of each model was about INR 25000/- (USD 350), excluding the cost of CT scan in our study. Safe surgery and shorter ICU stay offsets this additional cost. Some workers have used sandstone models to keep the cost low without compromising the quality.¹³ Our cases demonstrate that even in a resource-constrained country, this technology can be judiciously utilized. These models serve two other purposes, firstly it helps in preoperative counselling of a patient’s relatives, and secondly, it helps the residents and practitioners to improve their understanding of these lesions.^{15, 16, 17} The cost of these fringe advantages is beyond the scope of this study.

Conclusion

A small subset of complex CHDs need additional information regarding the 3D spatial anatomy for surgical correction. These patients benefit from 3D model reconstruction for surgical decision making. The commonest CHDs that need 3D model study are anomalies of ventriculoarterial relationships, namely conotruncal anomalies. This technology is a feasible option in terms of procedure and cost in the carefully selected group of patients to reduce the intraoperative surprises and optimize postoperative recovery.

Limitations of this study

The sample size of our study was small, and a larger sample size would be required to give further inputs and make the study statistically significant. It is recommended that this study may be taken as a pilot in this field, and a similar multicentric and multispecialty study with a larger sample size may be carried out.

Disclosure of competing interest

The authors have none to declare.