Three-dimensional print of a liver for preoperative planning in living donor liver transplantation

Authors


  • The study was supported by the Mikati Foundation Endowed Chair in Liver Diseases (Nizar N. Zein).

Abstract

The growing demand for liver transplantation and the concomitant scarcity of cadaveric livers have increased the need for living donor liver transplantation (LDLT). Ensuring the safety of donors and recipients is critical. The preoperative identification of the vascular and biliary tract anatomy with 3-dimensional (3D) printing may allow better preoperative surgical planning, avert unnecessary surgery in patients with potentially unsuitable anatomy, and thereby decrease the complications of liver transplant surgery. We developed a protocol and successfully 3D-printed synthetic livers (along with their complex networks of vascular and biliary structures) replicating the native livers of 6 patients: 3 living donors and 3 respective recipients who underwent LDLT. To our knowledge, these are the first complete 3D-printed livers. Using standardized preoperative, intraoperative, and postoperative assessments, we demonstrated identical anatomical and geometrical landmarks in the 3D-printed models and native livers. Liver Transpl 19:1304–1310, 2013. © 2013 AASLD.

Abbreviations
2D

2-dimensional

3D

3-dimensional

CT

computed tomography

LDLT

living donor liver transplantation

MRI

magnetic resonance imaging

STL

stereolithography

Three-dimensional (3D) printing is a process for making a solid 3D object of virtually any shape from a digital model. A 3D printer works as an ordinary office printer, but instead of placing a single layer of ink on paper, the machine lays down successive thin layers of a material to form a 3D object that replicates the original one.[1]

The growing demand for liver transplantation and the concomitant shortage of cadaveric livers have led to a rise in living donor liver transplantation (LDLT), in which resection of the right or left liver lobe is performed for the purpose of liver transplantation.[2] Living donors are healthy individuals, so ensuring their safety is of paramount importance. There have been a number of reported donor deaths worldwide and a substantial number of donor morbidities, so there is a need for measures to optimize donor safety.[3] Many of these morbidities are attributable to incomplete preoperative anatomical characterization of vascular and biliary structures and inaccurate estimates of the liver volume; these data are needed to determine the extent of the resection. This information provides a road map, and its accuracy has improved with the introduction of radiological software able to provide 3D visualization of liver structures.[4, 5]

3D imaging has the ability to better demonstrate the 3D relationships between vital vascular and biliary structures and the surrounding parenchyma in comparison with conventional computed tomography (CT) or magnetic resonance imaging (MRI). 3D imaging renders volumes as well, and these are essential to the surgical planning and execution process.[6-8] Despite these advantages, 3D imaging has its shortcomings, including the reality that 3D images are examined through a 2-dimensional (2D) computer screen. Additionally, the intraoperative application of preoperative 3D image findings is still difficult because of the absence of reliable liver surface markers corresponding to hepatic segmentation. Liver deformity after mobilization during surgery has been a major obstacle to the reliable use of preoperative 3D imaging data; hence, intraoperative modalities such as ultrasonography and cholangiography are often used for the localization of vessels and major biliary branches. Our attempt to produce a highly accurate 3D-printed model of a patient's liver is intended to overcome these obstacles and improve the safety of LDLT. The accurate anatomical reproduction of the donor liver represents progress in existing 3D imaging technology and may enhance patient outcomes.

The aim of this novel study was to establish anatomical precision and volumetric accuracy in 3D-printed models for donors and recipients undergoing LDLT.

PATIENTS AND METHODS

Patients

After receiving an approval from the Institutional Review Board at the Cleveland Clinic, we prospectively studied 3 consecutive patient pairs (3 donors and 3 recipients) who were evaluated and approved for LDLT at the Cleveland Clinic between November 2012 and May 2013. The donors and the recipients underwent standardized preoperative evaluations, which included contrast-enhanced CT and MRI examinations of the abdomen. These standardized images were required for an analysis of morphological characteristics and the vascular and biliary anatomy and for an evaluation of the hepatic parenchyma. Additionally, CT volumetric measurements of the entire donor liver and both its lobes were taken with CT postprocessing software (Medical Image Editor Thomas Lange, B.S., Deutsches Herzzentrum, Berlin, Germany). Using information from these imaging modalities, we obtained comparative data for the donor and recipient livers, including the volumes, lengths, widths, and heights of the whole livers and their lobes. Anatomical details, including the diameters of the main portal vein, left hepatic vein, right hepatic vein, and biliary tree, were also recorded. Demographic and clinical characteristics of the donors and recipients at the time of surgery (age, sex, ethnicity, weight, height, and body mass index) were also recorded.

3D Printing

The overall process of 3D printing included 3D image reconstruction, digital preparation, 3D printing, and postprinting finishing work. Anatomical structures (bile ducts, hepatic arteries, hepatic veins, and portal veins) were digitally segmented with MeVis-supplied (MeVis Medical Solutions AG) digital files from the optimal visualization phases of contrast-enhanced CT and MRI imaging. Subsequently, the geometry of each structure was exported to a mesh-type file [stereolithography (STL) file] to create a 3D geometry of that structure. Digital preparation of the STL files was accomplished with Magics software (Materialise) in order to (1) evaluate the intersecting vessels and the biliary structures originating from subtle time/position artifacts between CT/MRI imaging phases and then generate a nonoverlapping geometry, (2) actualize hollowed vascular and biliary structures within the liver parenchyma, (3) construct thin-walled external vessels and biliary structures to accompany the liver model, and (4) divide the liver mesh structure into graft and remnant components on the basis of the proposed surgical resection plane.

The final STL files generated via digital preparation were imported into a Connex 350 3D printer (Stratasys), and the 3D model was produced. Materials used for printing included TangoPlus/VeroClearPlus for liver parenchyma, TangoBlackPlus/VeroBlue for hepatic vein structures, and a TangoPlus/VeroClear blend (FLX9995-DM; Shore A hardness ∼92) for other external vessels (all materials by Stratasys). The postprinting phase involved cleaning of the support material that formed during the printing process with a water jet, selecting optimal color-coded dye staining for each vascular structure, and clear-coating the liver surfaces with methodologies that we had optimized over the past few months; this ensured maximum transparency of the model and vividness of the anatomical structures. Finally, the external vessels and the biliary structures were attached to the liver lobe via a permanent adhesive. This resulted in a life-sized, transparent 3D liver that not only replicated the real one but also allowed detailed visualization of vascular and biliary structures for optimal surgical planning.

The volumes of the native and 3D-printed right or left donor liver lobes were determined by means of liquid displacement with a special 6-L cylindrical stainless steel container that was open at the top and equipped with a stainless steel tube projecting from the top of its sidewall for overflow. The container was filled with saline to the level of the overflow tube. After the immersion of each 3D-printed liver in saline, the saline that escaped via the overflow tube was captured in a graduated cylinder. The volume of the displaced saline measured in the graduated cylinder corresponded to the volume of the measured part of the liver.

Intraoperative Data Acquisition

During surgery, hepatic vein mapping was performed with Doppler ultrasound, whereas the arterial and portal vein anatomy was assessed via surgical dissection. Additionally, after open cholecystectomy and ligation of the cystic duct, conventional intraoperative cholangiography was performed for bile duct anatomy. Anatomical information obtained from these intraoperative studies was used as endpoints for comparison with the 3D-printed model for accuracy. After proper identification of the vascular and biliary tracts of the donor's liver, the right or left liver lobe was resected for the purpose of liver transplantation.

A single individual with a standard laboratory scale determined the weight of the resected liver lobe before bench-table graft preparation. In the next step, measurements of the resected liver lobe, including the length, width, and height as well as the diameters of the portal vein, left hepatic vein, and right hepatic vein, were obtained by a single operator.

Similarly, measurements of the recipient's native liver were taken as soon as the liver was explanted, and its volume was determined via liquid displacement (as described previously). All measurements were taken by a single operator

Statistical Analysis

We used the results of intraoperative measurements as the gold standard for our study. The accuracy of the 3D-printed liver models was assessed through the comparison of their preoperatively obtained measurements to the gold standard. Data are presented as means and standard deviations. For each subject, the difference in each measure between the 3D print and the native liver was estimated; the mean differences and the corresponding 95% confidence intervals are reported. All analyses were performed with SAS 9.2 (SAS Institute, Cary, NC).

RESULTS

Case 1

The first case was a 42-year-old gentleman with cryptogenic liver cirrhosis and a 3-cm nodule in segment VI (according to Couinaud's classification). The nodule was nonenhancing according to contrast-enhanced CT and was consistent with a regenerative nodule. Figure 1A presents the 3D-printed liver and the native liver of the recipient. The 3D model accurately showed the regenerative nodule in segment VI. The geometric characteristics of the 2 livers (the 3D-printed and native recipient livers) were virtually identical, as shown in Table 1. The liquid displacement method was used for measuring the volume, and because the native recipient liver volume was considered the reference standard, the 3D liver model provided a more accurate volume measurement than preoperative volumetric CT (native liver volume = 1215 mL, 3D-printed liver volume = 1175 mL, CT volume = 1267 mL).

Figure 1.

(A) Preoperatively 3D-printed liver and actual explanted liver of a recipient (case 1). The arrows point to a regenerative nodule in segment VI in both the native liver and the 3D-printed liver. (B) Preoperatively 3D-printed right lobe and actual right lobe of a donor (case 1).

Table 1. Geometric Characteristics of 3D-Printed Liver Models and Corresponding Donor and Recipient Livers: Case 1
 Recipient's LiverDonor's Right Liver Lobe
3D PrintNative Right Liver Lobe3D PrintNative Right Liver Lobe
Volume (mL)11751215923988
Length (cm)21231313.5
Width (cm)13.312.515.515
Height (cm)10.51010.410.5
Portal vein diameter (mm)11.20.70.9
Right hepatic vein diameter (mm)0.40.50.40.4
Left hepatic vein diameter (mm)0.40.5

The donor was a 46-year-old brother who underwent right lobe hepatectomy (segments V-VIII) without the middle hepatic vein. The 3D-printed model and the native right liver lobe of the donor, as presented in Fig. 1B, appeared to be identical. Table 1 presents the geometric characteristics of the 3D-printed right liver lobe and the native right liver lobe. Intraoperative cholangiography was very difficult and time-consuming because of the very low insertion of the cystic duct (length = 4 cm), which caused the contrast to run into the duodenum. This anatomical variation was accurately shown in the 3D liver model, which allowed intraoperative characterization of the biliary anatomy when intraoperative cholangiography was suboptimal (Fig. 2). An accessory left hepatic duct draining directly into the common hepatic duct was also clearly noted in the 3D-printed model (Fig. 2).

Figure 2.

(A) Intraoperative cholangiogram (case 1). The arrow in panel-B points to an elongated cystic duct. (B) Preoperatively 3D-printed biliary tree.

Case 2

The second case was a 63-year-old gentleman with liver cirrhosis secondary to nonalcoholic steatohepatitis who underwent LDLT. Figure 3A presents the 3D-printed liver and the native liver of the recipient. The geometric characteristics of the 2 livers (the 3D-printed and native livers) were identical (Table 2). Using liquid displacement to measure the volume and considering the native recipient liver volume as the reference volume, we found that the 3D liver model provided a more accurate volume measurement than preoperative volumetric CT (native liver volume = 1195 mL, 3D-printed liver volume = 1235 mL, CT volume = 851 mL).

Figure 3.

(A) Preoperatively 3D-printed liver and actual explanted liver of a recipient (case 2). The long arrows point to the hepatic artery, the short arrows point to the hepatic vein, and the double arrows point to the portal vein. (B) Preoperatively 3D-printed right lobe and actual right lobe of a donor (case 2). The single arrows point to the hepatic artery, and the double arrows point to the portal vein.

Table 2. Geometric Characteristics of 3D-Printed Liver Models and Corresponding Donor and Recipient Livers: Case 2
 Recipient's LiverDonor's Right Liver Lobe
3D PrintNative Right Liver Lobe3D PrintNative Right Liver Lobe
Volume (mL)12351195715730
Length (cm)17.517.515.415.3
Width (cm)15.415.51010.1
Height (cm)7.57.59.59.5
Portal vein diameter (mm)1111.21211.8
Right hepatic vein diameter (mm)2.42.51.82
Left hepatic vein diameter (mm)2.12

The donor was a 27-year-old daughter who underwent right lobe hepatectomy (segments V-VIII) without the middle hepatic vein. Figure 3B presents the 3D-printed model of the native right liver lobe of the donor, which presented with geometric characteristics nearly identical to those of the native right lobe (Table 2). Additionally, the 3D model accurately showed the normal donor biliary anatomy in comparison with preoperative MRI and intraoperative cholangiography (Fig. 4). The biliary anatomy was conventional because the right posterior duct drained segments VI and VII, and the right anterior duct drained segments V and VIII.

Figure 4.

(A) Intraoperative cholangiogram (case 2). No biliary anomalies were noted. (B) Preoperatively 3D-printed biliary tree.

Case 3

The third case was a 63-year-old female with liver cirrhosis secondary to nonalcoholic steatohepatitis who underwent LDLT. Figure 5A presents a cross-section of the 3D-printed liver along with the corresponding cross-section of the native liver of the recipient. The 3D-printed cross-section of the liver accurately shows the orifices of the portal vein, hepatic vein, and hepatic artery and is consistent with the native liver cross-section. The geometric characteristics of these vascular structures appeared to be identical in the 3D-printed liver and the native liver of the recipient (Table 3). The 3D liver model provided a more accurate volume measurement than preoperative volumetric CT according to the postoperatively measured volume of the recipient's explanted liver (native liver volume = 1250 mL, 3D-printed liver volume = 1458 mL, CT volume = 1626 mL).

Figure 5.

(A) Cross-section of a preoperatively 3D-printed liver and corresponding cross-section of the actual explanted liver of a recipient (case 2). The arrows and the arrowheads point to the hepatic vein, and the dotted arrows point to the portal vein. (B) Preoperatively 3D-printed left lobe and actual left lobe of a donor (case 3). The arrows point to the caudate lobe in both the native liver and the 3D-printed liver.

Table 3. Geometric Characteristics of 3D-Printed Liver Models and Corresponding Donor and Recipient Livers: Case 3
 Recipient's LiverDonor's Left Liver Lobe
3D PrintNative Liver3D PrintNative Left Lobe
Volume (mL)14581250800755
Length (cm)272611.411
Width (cm)17.51714.614.2
Height (cm)6.361312.7
Portal vein diameter (mm)1010
Right hepatic vein diameter (mm)4.74.5
Left hepatic vein diameter (mm)5.65.412.612.3

The donor was a 43-year-old daughter who underwent left lateral segmentectomy for living liver donation. The 3D-printed model and the native left liver lobe of the donor are shown in Fig. 5B. Table 3 presents the geometric characteristics of the 3D-printed left liver lobe and the native left liver lobe. Supporting Table 1 presents the differences in each measure of the geometric characteristics between the 3D-printed and native livers.

DISCUSSION

In this study, using 3D printing technology, we have successfully produced prototype models of human livers based on patients' individual CT scans and MRI imaging. In our experience, 3D organ printing is a valuable tool for understanding the spatial relationships between vascular and biliary anatomical structures and ultimately for facilitating surgery and potentially minimizing intraoperative complications.

Through a direct comparative validation protocol, these models were shown to have a very high accuracy with mean dimensional errors of less than 4 mm for the entire model and less than 1.3 mm for vascular diameters (the portal vein and its main branches, the hepatic veins, and the hepatic artery). To our knowledge, this is the first time that human livers have been synthetically reproduced and validated against actual native livers at the time of surgery.

3D imaging is becoming an important clinical tool for the planning of complex surgeries and is considered superior to conventional 2D imaging for the visualization of anatomical structures.[4-8] A major disadvantage of 3D imaging is the visualization of images through a 2D computer screen, with which a true sense of depth is limited. Converting these images through 3D printing into a real 3D object with a real indication of depth that can be examined and altered in a way to be identical to the surgical cutting plane provides additional information that can influence surgical planning and potentially decrease complications. Additionally, having the physical model during surgical operations provides greater intuitive navigation for critical areas, and difficulties in orientation are overcome because these models can be physically positioned in the most desired fashion. Furthermore, the real-life intraoperative manipulation of the model simulates the changes in orientation experienced during the actual operation (eg, mobilization and retraction of the liver and exposure of the hepatic hilum), and this allows easy reorientation and identification of critical anatomical landmarks.

Finally, the transparency of the material used for the liver parenchyma in our 3D-printed models and the use of specific color codes for vascular and biliary structures provide detailed information about the vascular anatomy and the biliary anatomy, both of which are crucial for surgical outcomes. It may ultimately be possible to replace information provided by intraoperative studies (Doppler ultrasound and intraoperative cholangiography) with information provided by a 3D-printed model or to provide additional valuable information leading to reduced operative times and cold ischemia in liver transplantation. Because of the wide spectrum of materials that can be used in preparing these 3D copies of human livers, it is possible to use a soft material to simulate the mechanical properties of actual liver tissue so that models can be used to rehearse liver resection in LDLT or tumor resection in hepatobiliary and transplant surgery.

Although we have presented potential incremental progress in accurate preoperative and intraoperative surgical planning beyond conventional and 3D imaging, certain limitations should be noted. 3D-printed physical models are based on imaging and accordingly will be prone to imaging errors. Improved imaging resolution will subsequently improve the accuracy of these 3D-printed models. However, according to the present report, the level of accuracy rendered by these 3D models seems to be highly acceptable with small margins of error in comparison with the explanted native livers at the time of surgery. Another potential limitation is the need to control a number of parameters, including the type, quality, and properties of materials used, in order to achieve a reliable and reproducible model. Finally, the issue of cost is a constant concern when new technologies are being introduced. The production of a single 3D model requires approximately 25 to 40 hours of labor and the cost of materials in addition to the cost of imaging alone. Our group is working on a number of initiatives that will likely improve the efficiency and lower the cost of 3D-printed liver replicas.

In conclusion, we have presented the successful reproduction of human livers via 3D printing technology. These highly accurate models provide practical and hands-on tools and have a number of possibly unique applications for surgical planning and medical education.

Ancillary