3D-Printed Models for Multidisciplinary Discussion of Congenital Heart Diseases

,


Introduction
Congenital heart defects (CHDs) are complex and widely variable anatomic lesions that present serious diagnostic and therapeutic challenges.Tree-dimensional (3D) printing allows understanding of the 3D orientation and spatial relationship of cardiovascular structures in CHDs.3D-printed anatomic models have had various applications in trainee education, surgical/interventional planning, patient/family education, and communication in medical practice [1][2][3][4].However, due to their challenging nature, there are still knowledge gaps and limited data in this area in terms of randomized studies and comparative research on the outcomes of 3D printing, especially considering the variety of available software, hardware, techniques, and printing materials [5].Diferent centers have reported diferent experiences and practices in 3D printing.Terefore, we prospectively investigated the clinical value and feasibility of a 3D-printedpatient-specifc model for multidisciplinary discussions of various complex CHDs in a single tertiary center.

Materials and Methods
From August 2019 to April 2021, we conducted an openlabel prospective pilot study, using 20 3D-printed models of 19 patients with complex CHDs during 20 multidisciplinary discussions among 8-14 pediatric cardiologists and cardiothoracic surgeons.Te selection of cases for 3D printing was decided at a multidisciplinary meeting.After obtaining informed consent from the patients and/or their parents, depending on the participant's age, a cardiac computed tomography (CT) scan was used to generate a DICOM fle.Segmentation and postprocessing of the cardiovascular structure were performed, and standard tessellation language fles were generated using commercially available software (MEDIP PRO v2.0.0.0., MEDICAL IP, Seoul, Korea) (Figure 1).Patient-specifc3D-printed models were produced in two types: "blood pool model" and "hollow model," which consisted of the lining around the blood pool model with a wall thickness of 1-2 mm, meticulously representing the intracardiac anatomy (Figure 1, right bottom) [6].Using the 3D-printed model, we discussed the management plan in a multidisciplinary meeting.All participating cardiologists and cardiac surgeons were given a questionnaire, to which they had to respond on a scale of 0-10, with 10 indicating the highest score.Te questions in the questionnaire surveyed the efect of the model in understanding the 3D orientation of the cardiovascular anatomy, designing a surgical plan, predicting surgical complications, facilitating multidisciplinary discussions, and communication, how it changes comfort and confdence in management, and if there was a change in the management plan after the use of the 3D-printed model (Supplementary Table 1).Patient demographic and clinical data were extracted from electronic medical records.
An intraclass correlation coefcient and Bland-Altman plot were obtained for 20 printed models to illustrate the agreement between the phantom CT of the 3D-printed model and the cardiac CT (Supplementary Figure 1).Te sizes of the ventricular septal defect (VSD), aorta, superior vena cava, inferior vena cava, and pulmonary arteries were measured and compared between the 3D-printed model and phantom CT.
Te study was approved by the Institutional Review Board of the Seoul National University Hospital (No. 1904-031-1024).Written informed consent for participation from the patients and/or their parents was obtained before producing the 3D model.Moreover, informed consent was obtained from the clinicians who participated in the discussion and answered the questionnaire.

Statistical Analysis
Continuous and ordinal variables were expressed as means ± standard deviations or medians and ranges, as appropriate.Pre-and post-3D model confdence and comfort scores were compared using the Wilcoxon signedrank test.Categorical variables are expressed as frequencies (percentages).Te chi-square and Fisher's exact tests were used to compare categorical data between two or three groups.Statistical analyses were performed using SPSS software (version 23.0,IBM Corporation, Armonk, NY, USA).Statistical signifcance was established at p < 0.05.

Results
During 20 multidisciplinary discussions, a total of 212 questionnaires were completed.Te median age and weight of the patients for whom 3D models were printed were 0.8 years (range, 5 days to 43 years) and 9.6 kg (range, 2.8 to 54 kg), respectively.Te clinical diagnosis and reason for printing a 3D model for each patient are summarized in Table 1.Tere were 12 patients with a double outlet of the right ventricle (DORV), which was the most common underlying disease.In those patients, the 3D models were used in the discussion that led to the selection of the treatment plan between biventricular repairs versus Fontan palliation.Tree patients had multiple or unusual locations of VSDs.In two patients with hypoplastic left heart disease and bilateral     2).A neonatal male patient (Table 1, models 6 and 13) with a superoinferior ventricle, DORV, subaortic VSD, rightsided atrial appendage juxtaposition, a nearly single atrium, and mesocardia in situs solitus (Videos 1 and 2 and Figure 3).Te VSD plane was not fully understood via CT and echocardiography.Biventricular repair was initially impossible because VSD bafing would likely cause subpulmonic stenosis, and there was no space for conduit placement between the right ventricle and the pulmonary artery.Although the patient had progressive congestion and mild cardiomegaly, we decided to closely monitor him and wait for biventricular repair.Te pulmonary artery had grown with age, and at 11 months of age, the 3D-printed model suggested a possibility of VSD bafing without risking subpulmonic obstruction.Consequently, biventricular repair with VSD bafing, ASD patch partitioning, and right ventricular outfow tract widening was performed.

Multidisciplinary Discussion.
According to the answers to the questionnaire, the 3D-printed model accurately represented cardiac structures (9.4 ± 0.7), helped clinicians understand spatial orientation (mean rating score, 9.4 ± 1.1), allowed for easy and quick communication among coworkers (9.4 ± 0.9 and 9.2 ± 1.1, respectively), aided in the prediction of surgical complications (9.0 ± 1.1), and provided additional information over conventional imaging (9.2 ± 0.4) (Supplementary Figure 3).Comfort and confdence in the surgical plan signifcantly increased after using the 3D-printed model (pre, 6.2 ± 1.6 versus post, 9.2 ± 0.9, p < 0.001 and pre, 6.3 ± 1.6 versus post, 9.2 ± 0.8, p < 0.001, respectively) (Figure 4).Pediatric cardiologists and cardiac surgeons did not difer signifcantly in whether 3D-printed models accurately displayed the cardiac structure, helped to understand the 3D orientation, or simplifed communication between clinicians, or in their preoperative /postoperative comfort and confdence after using a 3D-printed model.However, they difered regarding whether the model shortened the discussion; the median rating score awarded by cardiologists and cardiac surgeons were 9.37 ± 0.87 and 9.03 ± 1.32, respectively (p � 0.033).When prompted to provide additional comments, respondents mentioned limitations, such as that the 3D-printed model did not satisfactorily represent the valve, and that the simulation was inaccurate because the material with which the 3D model was printed difered from actual heart tissue.

Discussion
Our study demonstrated that patient-specifc3D-printed models accurately represented the cardiac structure, except for the cardiac valve, exhibiting good correlation between chest CT and phantom CT of the models (intraclass correlation coefcient of 0.996 and no signifcant bias according to the Bland-Altman plot).Tey also provided insight into the 3D spatial orientation of the defects and helped physicians in their decision-making on the surgical plan and in the prediction of surgical complications.Te 3D model facilitated communication and reaching an agreement in multidisciplinary discussions.Comfort and confdence in the surgical plan increased signifcantly with the 3D model, which illustrates the importance of this tool in preparation for surgery.Furthermore, pediatric cardiologists and cardiac surgeons did not difer in the degree to which they felt the models facilitated decision-making, communication, and their understanding of the 3D anatomic orientation of the defects.Although they difered in their opinions regarding the degree to which the models shortened discussions, both mean scores were more than 9.0 (9.37 ± 0.87 and 9.03 ± 1.32, respectively).Taken together, both pediatric cardiologists and cardiac surgeons found the 3D models helpful in various ways, and cardiologists particularly thought that the 3D model shortened the discussion on patient management for CHDs.
Randomized trials or comparative studies on patients with complex CHDs are challenging to perform because they are rare and extremely heterogeneous.Patients difer in the anatomy, combinations of VSD locations, relationship between the great vessels and ventricles, and sizes of the ventricles, VSDs, atria, great arteries, and cardiac chambers.Although difcult to quantify, a patient-specifc3D-printed cardiovascular model and surgical/interventional simulation may decrease pump and procedure times, decrease complications, and improve surgical/interventional outcomes.
3D-printed models have been used in patients with CHDs since the early 2000s, and patient-specifc 3D models have become more widely used in surgical planning, surgical/percutaneous interventional simulation, and patient/ family education and communications in the last decade [5].Patient-specifc 3D models have exhibited good agreement with CT scans and magnetic resonance images, while the technique and its application have advanced and evolved into augmented and virtual reality [2,6].Models help in surgical decision-making but can also lead to changes in the surgical plan [7,8].For example, cross-sectional images alone are insufcient for assessment of the feasibility of biventricular repairs by VSD bafing, as the intracardiac space is limited, and the conduit is positioned between the right ventricle and pulmonary artery, resulting in obstruction.Tis is especially true in patients with a DORV and remote or subpulmonic VSDs, and in those with an unusual superoinferior relationship of the ventricles.However, the 3D model allows clinicians to visualize the anatomical relationship between cardiovascular structures, which may lead to a modifcation of the surgical method to be used [8,9].Decision-making regarding Fontan surgery versus biventricular repair requires meticulous consideration, particularly in patients who are suboptimal candidates for biventricular repair, as the decision directly impacts the patient's long-term prognosis.Te usefulness of 3D anatomic models in such complex and controversial decisionmaking is indisputable [10].
In our study, the 3D-printed model also helped the surgical team to select a surgical method and allowed simulation a challenging case that required VSD extension after a Rastelli operation for complete transposition of the great arteries, a VSD, and pulmonary stenosis (Figure 2).In patients with pulmonary atresia and MAPCAs, the 3D model helped clarify the anatomic and spatial relationship of the MAPCAs, native pulmonary artery, and airways.Te use and applications of 3D modeling and printing have increased and are constantly evolving.It has advanced to the point where it is being used in hands-on surgical training and in computer-aided sterilizable templates for bafes and patches for complex surgical procedures [10].However, its application and utility difer greatly from country to country and institution to institution, which may be because of difering patient groups and fnancial/insurance contexts.
Te process of 3D printing is time-consuming, laborintensive, and expensive, which are obstacles to using 3D models.In our experience, creating a 3D model takes at least seven days, as the segmentation requires manual editing by CHD medical experts, communication between medical experts and software experts, and further refnement with computer-aided design software.Furthermore, the valve and subvalvular apparatus cannot currently be accurately reproduced with a 3D-printed model.Te texture of the model difers from that of the real heart, and the models are more easily cut or torn than real cardiac tissue.Although Figure 3: A neonatal patient with a superoinferior ventricle, DORV, subaortic VSD, right-sided juxtaposition of the atrial appendages, and mesocardia in situs solitus.Te main pulmonary trunk was located posteroinferiorly, the subvalvular infundibulum exhibited mild to moderate stenosis, and the main and branch pulmonary arteries were elongated.Te entire heart was rotated clockwise and inferiorly along the axis from the apex to the base.Te LV apex was pointing in the subxiphoid direction, where apical beating was visible (a, d).Te 3Dprinted model clearly showed the spatial relationship of the sternum, ventricle, atrium, and great vessels, as well as the intracardiac ventriculoarterial relationship in this complex heart disease (b, c, e, f ).Ao, aorta; AP, anteroposterior; DORV, double outlet of the right ventricle; IVC, inferior vena cava; LAA, left atrial appendage; LV, left ventricle; MPA, main pulmonary artery; PA, pulmonary artery; RAA, right atrial appendage; RV, right ventricle; RVOT, RV outfow tract; TV, tricuspid valve; VSD, ventricular septal defect.software, hardware, and 3D-printing material have improved during the study period, further study and improvements are required.

Conclusion
Patient-specifc3D-printed models improved the understanding of complex CHDs and facilitated multidisciplinary discussions and surgical decision-making in our study.However, a limitation of this study is that the measured outcomes were based on subjective reports.Teir wider use will require further study and improvements in terms of the cost and time for necessary for their production, as well as the materials from which they are constructed.
An anterior oblique view on echocardiography of model 6 shows the left pulmonary artery with stenosis.(Supplementary Materials)

Figure 1 :
Figure 1: Construction of the three-dimensionally-printed model.

Figure 2 :
Figure 2: A 43-year-old patient with severe subaortic stenosis.Tis patient had complete transposition of the great arteries, a VSD, and pulmonary stenosis, and underwent VSD bafing and a Rastelli operation at 13 years of age.However, the previous VSD became restrictive, and the subaortic stenosis gradually progressed.(a) Left ventricular angiography revealing subaortic stenosis.(b) Upon echocardiography in the high parasternal modifed long-axis view, the peak velocity at the subaortic level was measured as 5 m/s.(c) Te left lateral cutting plane view after 3D segmentation using the software.Te white arrow indicates the expected location of the conduction system.(d) Te patient-specifc3D-printed hollow model.Te thick arrow indicates the restrictive VSD.VSD widening was planned using a transaortic approach (arrow).LV, left ventricle; VSD, ventricular septal defect.

Table 1 :
Patients' diagnosis and reasons for 3D-printed model.
BCPS, bidirectional cavopulmonary shunt; cAVSD, complete atrioventricular septal defect; DORV, double outlet of the right ventricle; fSV, functional single ventricle; HLHS, hypoplastic left heart syndrome; MAPCA, major aortopulmonary collateral arteries; PA, pulmonary atresia; PAB, pulmonary artery banding; PS, pulmonary stenosis; TGA, transposition of great arteries; TOF, tetralogy of Fallot; VSD, ventricular septal defect.Printed Models.A 43-year-old female patient (Table1, model 1) with complete transposition of the great arteries, a VSD, and pulmonary stenosis had severe subaortic stenosis (pressure gradient between the aorta and left ventricle � 73 mmHg, left ventricular pressure � 196 mmHg) due to a restrictive VSD after a Rastelli operation at the age of 13 years.Her NYHA functional class was II-III, and the stenosis had gradually progressed.Widening of the VSD was required, but there was a high risk of heart block.We had multiple discussions with pediatric cardiologists, cardiothoracic surgeons, radiologists, and pathologists, using a 3Dprinted model, on how to extend the VSD without causing heart block.Eventually, the patient underwent subaortic muscle resection without complications.A transaortic approach and posterior muscle resection were performed to avoid conduction bundle injury.Te subaortic stenosis was relieved, and the NYHA class improved to I-II (Figure