Patient-Specific Coronary Artery Bypass Graft 3D Printing: Implications for Procedural Planning in Complex Percutaneous Coronary Interventions

Background. Three-dimensional (3D) printing technology has seen tremendous growth in augmenting didactics, research, and preprocedural planning with structural heart procedures. Limited investigative efforts have been made in other areas of the cardiovascular spectrum. 3D-printed models (PMs) of anatomically complex coronary artery bypass graft (CABG) patients from coronary computed tomography angiography (CCTA) have implications for adaptive learning and preprocedural planning. Methods. Five patients with CCTA who underwent subsequent coronary angiography were 3D printed for retrospective comparisons. Standard slicer software was used to create a computer-aided image of the ascending aorta, native coronary arteries, bypass grafts, aortic arch, and great vessels and 3D printed using polylactic acid filament. The models were painted with acrylic paint to highlight anatomical features and comparison was made with coronary angiography and 3D-CTA images. Results. All occluded vein grafts, left and right internal mammary artery (IMA) grafts, patent saphenous vein grafts, along with distal graft anastomotic sites, were accurately 3D printed. In cases with chronic total occlusions (CTOs), ambiguous ostial caps, mid or distal vessel chronic occlusions, and occlusions seen as CTOs on coronary angiography were 3D printed showing either distal vessel reconstitution via collaterals or complete arterial filling seen in a setting of calcification, microchannels, and collateral flow. Lastly, 3D printing of the aortic root and great vessels allowed for better appreciation of vessel tortuosity to aid in the cannulation of IMA grafts and optimizing engagement with diagnostic and guiding catheters. Conclusions. 3D printing of anatomically complex CABG patients has the potential to assist with preprocedural planning and operator understanding of complex coronary anatomy.

J INVASIVE CARDIOL 2021;33(8):E592-E603.

Key words: CAD, CABG, chronic total occlusion, complications, PCI

Three-dimensional (3D) printing technology has been successfully utilized within numerous medical specialties for both research and educational purposes, such as pharmaceutics for drug discovery, stem cell for tissue and organ regeneration, prosthetics and implant creation, and anatomical models for procedural planning.¹ There has been, however, limited investigative effort made toward its potential in the diagnosis and treatment of coronary artery disease (CAD). Coronary artery bypass graft (CABG) surgery patients represent a complex subset, with an increased risk of procedure-related complications during diagnostic and percutaneous coronary intervention (PCI) related to requirement of multiple catheters for engagement associated with increased procedural time and contrast volume. Great vessel tortuosity can further add to procedural complexity, and these anatomical complexities translate to higher rates of procedure-related stroke. The learning curve for new operators to become confident in their diagnostic and PCI capabilities in CABG patients is also steeper due to the aforementioned factors. A coronary computed tomography angiography (CCTA) is often the next step to delineate complex anatomy in cases where a diagnostic or interventional procedure is unable to be completed safely in a timely and contrast-controlled manner from occluded or angulated origin of grafts, or ambiguous caps in PCI for chronic total occlusion (CTO). The addition of 3D-reconstructed computed tomography angiography (3D-CTA) allows further improved understanding of a two-dimensional (2D) representation of 3D anatomy. 3D-printed models (3D-PMs) can bridge the gap between a virtual image to that of a real-life model for enhanced appreciation of anatomy to guide learning and procedural planning.   

Patients with complex coronary bypass anatomy requiring a CCTA for various reasons who also underwent coronary angiography at a single center were queried for retrospective comparison. Serial images from CCTA were downloaded into the commercially available 3D slicer software (Surgical Planning Laboratory) with reconstruction of a 3D image and slicing of relevant anatomy, including ascending aorta, native coronary arteries, bypass grafts, and in some cases, aortic arch and great vessels. A computer-aided image in the form of standard tessellation language (STL) file was created and the .STL file was uploaded into the Ultimaker 3 software (Create Education Limited). The models were 3D printed using Ender 3 Pro (Creality 3D) printers, already available within the institution. Fused deposition modeling, a method that heats thermoplastic filament and extrudes it layer by layer, was utilized using polylactic acid (PLA) filament.² The 3D-PMs were then painted using acrylic paint to enhance anatomic features for clinical analysis. Comparisons of the 3D-PMs were made with coronary angiography and 3D-reconstructed CTA images to determine clinical value in procedural planning. 

Five patients who met the inclusion criteria were reviewed and 3D-PMs were produced. The clinical history and indication for CCTA of all 5 patients are listed in Table 1. The coronary angiography operator’s interpretation and the CTA reader’s interpretation are listed in Table 2 and Table 3 Part 1 and Part 2, respectively. The anatomy for each of the individual cases is described below along with the 3D-PM features.

Case 1 (Figure 1 and Figure 2): 

a. Buttons of the occluded vein bypass grafts to obtuse marginal (OM) branch, ramus intermedius (RI) and right coronary artery (RCA) are readily identified (1Aa-1Ca) and are noted to be anatomically correct when compared to coronary angiography (1Bb). The operator chose not to selectively engage the vein graft to RCA to avoid renal failure from excessive contrast, as it was reported to be occluded from prior angiogram, which is confirmed on CTA and 3D-PM. The occluded vein graft to RCA can be clearly seen below and slightly posterior to the RI occluded vein graft (1Aa).

b. The patent left internal mammary artery (IMA) to mid-segment left anterior descending (LAD) coronary artery is seen (1Ab, 1Bc, 1Cc) along with distal anastamotic site and diffusely diseased native distal LAD (1Ab, 1Bc-1Bd, 1Cc).

c. Ambiguous ostial RI chronic total occlusion (CTO) cap difficult to discern even on 3D-CTA (1Cb) is more easily identified on the 3D-PM (1Ab-1Ac) and precisely identifies the take-off and angulation, along with the distal vessel course from collateral filling (1Ab-1Ac, 1Bf, 1Cb). 

d. Proximal RCA-CTO clinical course: proximal RCA-CTO can be seen on initial diagnostic angiogram (1Ba) with left-to-right epicardial collaterals from distal left circumflex (LCx) (1Be). Prior to the CCTA, the patient had a stent placed from his left main into the LCx, which further improved left-to-right epicardial collateral flow to the RCA from the distal LCx (2B, 2C). An ostial RI-CTO PCI was also attempted but failed due to ambiguity of the cap (2D). Next, CTO-PCI was attempted for his RCA, first with retrograde wire escalation (RWE) from the LCx epicardial collaterals (2E), failing but leading to improvement in left-to-right collateral flow (2F). Failed antegrade wire escalation (AWE) attempt of RCA-CTO and subsequent angioplasty of severely diseased proximal RCA are shown in 2H. The final angiogram at completion of the procedure still showed an occluded proximal to mid RCA with improved antegrade microchannel and collateral flow (2I). The CCTA was performed a few weeks after these interventions and unexpectedly showed complete RCA filling past the distal bifurcation (noted on 3D-CTA and 3D-PM). This finding was likely due to the presence of severe vessel calcification along with improved left-to-right and right-to-right collateral flow and from creation of microchannels (subintimal and intraluminal) from the prior failed CTO-PCI attempt.  

Case 2 (Figure 3): 

a. Occluded vein graft buttons to diagonal branch and RCA (inferior and posterior to that of the diagonal branch) can be seen (3Aa). 

b. Vein graft to OM1 is patent with suspected mid vein graft thrombus (3Bc), confirmed on CTA and 3D-PM (3Ab). Vein graft to OM2 branch is patent (3Ab and 3Bd).

c. The proximal LAD has critical disease and is occluded at the mid segment. The left IMA is atretic, as seen on coronary angiogram with a severe distal anastamotic site stenosis restricting flow to distal LAD (3Ba, 3Be). Due to the complex flow dynamics of proximal LAD disease and limited antegrade flow from atretic LIMA graft with anastamotic stenosis into the distal LAD, neither the LAD nor the LIMA graft could be 3D printed and is seen as an occlusion proximally on the 3D-PM (3Ac). The LCx is occluded proximally (3Ac, 3Ba).

d. The RCA is severely diseased proximally and then occluded at the mid segment, as shown (3Aa, 3Bb).

Case 3 (Figure 4 and Figure 5): 

a. Patent left main coronary artery (4Aa, 4Bb).

b. Vein graft to mid LAD is widely patent without any significant stenosis and the anastamotic site can be clearly seen at the mid segment of the LAD with patent native distal vessel (4Ab, 4Bd, 4Ca). 

c. Vein graft to OM1 is widely patent and the anastamotic site can be clearly seen; native distal OM is patent without significant disease (4Ac, 4Be, 4Ca). The atrioventricular groove/LCx is also seen (4Ac, 4Bb). 

d. The native RCA, although chronically occluded at the ostium on diagnostic angiogram (4Ba, 5A), is noted to fill until the distal RCA on 3D-CTA and 3D-PM (4Aa, 4Cb-4Cc, 5E-5G). As seen in case #1, this unexpected finding is likely from native vessel calcification along with robust left-to-right epicardial collaterals from the distal LCx allowing retrograde filling to the proximal RCA (5C-5D), septal collaterals from LAD vein graft (5B), and antegrade proximal vessel microchannel filling (5A).

Case 4 (Figure 6): 

a. Patent vein graft to diagonal branch can be seen with anastamosis to the diagonal branch that is severely diseased proximally (6Aa, 6Bc, 6Ca). Occluded vein graft to the OM is noted superior and anterior to the diagonal vein graft (6Aa). 

b. The vein graft to the right posterior descending artery (PDA) has moderate proximal disease and graft angulation at the distal graft anastamotic site (6Ab, 6Bd, 6Ca-6Cc). Native RCA with severe proximal disease is shown (Figure 6Ab, 6Ba, 6Cc).

c. The native LCx has severe disease proximally and is occluded past the mid segment (6Ac).  

d. The native LAD is critically diseased at the proximal segment (6Aa, 6Bb). The left IMA graft is patent and the anastamotic site to the mid LAD can be seen, the distal LAD past the distal anastamotic site is severely diseased (6Ab-6Ac, 6Be, 6Ca).

Case 5 (Figure 7): 

a. Ostial RCA vein graft stent thrombosis and an occluded native proximal RCA with distal vessel reconstitution and some intermittent flow mid-vessel due to microchannel flow (7Aa, 7Ba-7Bb, 7Ca).

b. Occluded vein graft stump to OM branch (superior and anterior to RCA vein graft), which could not be engaged on coronary angiography despite use of multiple diagnostic catheter, but is clearly visualized on 3D-PM (7Ab).

c. Patent right IMA graft to mid segment LAD and proximal native LAD chronic occlusion (7Ab, 7Bc). Patent left IMA to diagonal branch with angiographic and 3D-CTA correlate (7Ab, 7Bc-7Bd, 7Cc). 

d. Patent left IMA to diagonal branch (7Ac, 7Be, 7Ca). 

e. Severely diseased left main coronary artery (7Ac).

3D printing in structural heart procedures. 3D printing technology has been utilized effectively clinically, especially with structural heart disease procedures, as it precisely and accurately replicates complex anatomy in order to improve understanding but also to anticipate procedure-related complications, thus impacting vital intraprocedural decisions. However, there has been limited effort to evaluate its potential impact with planning and treatment of cardiovascular disease, which still remains the number 1 cause of death worldwide. Current applications of 3D-PM include: replication of left atrial appendage (LAA) anatomy to aid with device selection for LAA occluder devices;^3,4 aortic models for transcatheter aortic valve replacement (TAVR) procedures to predict coronary artery occlusion, paravalvular leaks, replication of flow dynamics, and assessment of geometric orifice area and calcium deposition at leaflets;^5,6 printing of mitral valve anatomy for transcatheter mitral valve procedures to evaluate for left ventricular outflow tract obstruction and optimal valve seating for a successful outcome;⁷ and atrial septal models for transcatheter atrial septal closure procedures to assess for adequate atrial rim tissue available for device deployment.⁸ 

Experience with 3D printing of coronary arteries. 3D-PMs have been helpful in portraying coronary anatomy and being complementary to 3D-CTA when a survey of participants was conducted with 3D printing of 8 anomalous coronary arteries from CTA.⁹ In another study, patient-specific 3D-PM of a segment of coronary artery from coronary angiography and optical coherence tomography (OCT) used to generate a computer-aided image found it to be highly accurate with respect to diameter of vascular stenosis.¹⁰ This holds important implications, as it may allow for patient-specific stent printing and sizing for tailored therapy. An attempt at assessment of hemodynamic significance has also been made by Yang et al, where 3D-PM of 10 left coronary artery stenoses showed that spatial characteristics (curvature of culprit vessel x angle of culprit vessel to upstream parent branch) was an index of hemodynamic significance and maximum velocity reduction post stenosis correlated with fractional flow reserve (FFR) CT. The study also concluded that 3D-PM more accurately represented coronary anatomy than virtual 3D models, providing enhanced visual and sensory perception not afforded by virtual 3D.¹¹ Hemodynamic significance of coronary stenoses in the form of FFR was further tested as benchtop FFR (B-FFR) of 3D-printed coronary models and compared with that obtained with CT-FFR by Sommer et al. Hyperemia was simulated in 52 models as coronary flow rate of 500 mL min-1 and the CT-FFR and B-FFR were found to be similar in the 2 arms, with results showing area under the curve (AUC) to be 0.80 (95% confidence interval [CI], 0.70–0.87) and 0.81 (95% CI, 0.64–0.91), respectively.¹² A recent case report illustrated use of 2D angiography to 3D print from a 3D volume rendering of the left coronary artery, which was then connected to a custom-made simulator. The 3D-PM was used to treat complex in-stent restenosis of ostial LCx with OCT guidance prior to the actual procedure.¹³ These studies highlight the importance of 3D-PM of the coronary anatomy in educational training, patient-specific treatment for coronary disease, and opportunities for research, expansion of knowledge, and guidance of therapeutic interventions. As of yet, there have not been any attempts to ascertain feasibility or clinical utility of 3D-PM of CABGs in complex coronary interventions. 

Clinical risks with CABG angiography. The incidence of CABG is approximately 8% in all patients undergoing coronary angiography, as seen in a hospital record of 5993 patients.¹⁴ It is well established that patients with history of CABG represent a complex subset of patients who are often older, with higher prevalence of comorbidities including diabetes and peripheral arterial disease. A retrospective study of 681 patients with prior CABG undergoing coronary angiography found the single strongest predictor of major adverse cardiovascular events (MACE) and vascular complications to be procedural time, with each additional 10 minutes adding 1.261-fold increased risk of MACE, and a 1.826-fold increased risk of vascular complications.¹⁵ Additionally, the presence of concomitant chronic kidney disease can further complicate the procedure, especially in cases where graft anatomy may be unknown, further increasing procedure time and contrast volume. A retrospective study of 392 diagnostic coronary angiograms comparing patients with prior CABG with known vs unknown CABG anatomy found that those with unknown anatomy had higher amount of contrast use (189 ± 7 mL vs 158 ± 4 mL), longer fluoroscopy times, and more diagnostic catheter use.¹⁶ Furthermore, stroke risk is higher in patients undergoing coronary angiogram with prior CABG. Patient-related factors known to increase risk of stroke with coronary angiography include older age, hypertension, diabetes, prior history of stroke, renal failure, heart failure, and severity of CAD including triple-vessel disease.^17-20 Procedure-related factors known to increase stroke risk include emergent catheterization, longer procedural time, greater contrast use, interventions at bypass grafts, retrograde catheterization of left ventricle with aortic stenosis, use of intra-aortic balloon pump, and presence of coronary artery thrombus.^21-24 Thus, patients with a history of CABG and/or any of the above risk factors are already at a higher risk of periprocedural related stroke. Thereby, 3D-PM of coronary anatomy may be a useful strategy to understand complex anatomy or facilitate cannulation of bypass grafts in order to mitigate some of these risks. 

Role of imaging and benefits of 3D printing of CABG patients. The diagnostic accuracy for assessment of ≥50% stenosis in venous or arterial grafts by CTA is high, with a 100% sensitivity and specificity in one study.²⁵ In our study, all occluded and diseased grafts were accurately printed on all five 3D-PMs. However, a known limitation of CTA is its ability to differentiate from very high-grade stenosis to that from CTO, which has been related to partial volume effect and lower spatial resolution of CTA to that of coronary angiogram.²⁶ Hence, a confirmation with coronary angiogram is usually required when a total occlusion is interpreted on CTA. Nonetheless, CTA has the ability to identify calcification, vessel tortuosity, occlusion length, distal vessel anatomy past occlusion, and ostial CTO stumps.²⁷ In one report, CTA had the ability to identify distal vessel and route for CTO-PCI in 68% vs 18% with use of coronary angiogram alone.²⁸ Current published data also show that ostial CTOs are associated with longer fluoroscopy times, more frequent retrograde approach for recanalization, and lower success rates than non-ostial occlusions.²⁹ Hence, 3D-PM could assist and perhaps tailor the hybrid algorithm in CTO involving CABG patients, as ostial CTO could be better approached than current practices³⁰ by identification of chronic ostial cap occlusion, appreciation of proximal cap angulation, identification of tapered or blunt cap, and distal vessel course, all of which can alleviate cap ambiguity. This is illustrated in Case #1, where the operator was unable to engage the RI ostial cap by coronary angiogram alone due to ambiguity. The ostial CTO cap and distal vessel were then accurately delineated with the 3D-PM. Additionally, Case #1 and Case #3, which were shown to be CTOs on coronary angiogram, were demonstrated to almost entirely fill on the 3D-PM. This is likely due to vessel calcification, improved collateral flow, antegrade microchannel filling, and in Case #1 due to prior failed revascularization attempt with proximal vessel PCI also contributing to improved flow allowing the vessel to appear patent on CCTA. Thus, enhanced visual and conceptual perception offered by 3D-PM vs its 3D-CTA counterpart holds the potential to impact procedural planning and improve CTO-PCI success rates in this subset. 

In summary, 3D-PM of complex coronary anatomy in CABG patients has several advantages, including: decreasing procedural time and contrast use in difficult-to-find bypass grafts; preprocedural planning in CTO-PCI with assessment of ambiguous proximal or ostial CTO caps, morphology, vessel course, and distal vessel anatomy; visualization of graft anastomotic site; preprocedural planning where vein graft is utilized for retrograde conduit in CTO-PCI; and navigation of aortic and great vessel tortuosity when engaging angulated venous or arterial graft takeoffs. 3D-PM can bridge the gap between a virtual image and real-life model, providing superior understanding of complex anatomy in CABG patients.