- Email: [email protected]
The accuracy of patient specific implant prebented with 3D-printed rapid prototype model for orbital wall reconstruction
Journal of Cranio-Maxillo-Facial Surgery 45 (2017) 928e936
Contents lists available at ScienceDirect
Journal of Cranio-Maxillo-Facial Surgery journal homepage: www.jcmfs.com
The accuracy of patient specific implant prebented with 3D-printed rapid prototype model for orbital wall reconstruction Young Chul Kim a, Woo Shik Jeong a, Tae-kyung Park b, Jong Woo Choi a, Kyung S. Koh a, Tae Suk Oh a, * a b
Department of Plastic Surgery, Asan Medical Center, University of Ulsan College of Medicine, Seoul, Republic of Korea Department of Emergency, The Newcastle Upon Tyne Hospitals, Newcastle Upon Tyne, United Kingdom
a r t i c l e i n f o
a b s t r a c t
Article history: Paper received 22 August 2016 Accepted 20 March 2017 Available online 27 March 2017
Background: This study evaluated the accuracy of blow out fracture reduction using 3D-printed rapid prototyping (RP) skull modeling. Patients and methods: Retrospective review was performed for 82 patients who underwent posttraumatic orbital wall fracture reduction between 2012 and 2014. Patients were divided into two groups according to the use of 3D-printed RP skull model reproduced by mirroring technique, onto which a titanium mesh was anatomically molded. Using computed tomographic scans, the areas of preand post-operative orbital wall defect, the layout angles of the titanium mesh, and the gap lengths between the implant and fracture margin were compared between the two groups. Results: Of the 82 patients identified, 46 and 36 were diagnosed with medial and inferior orbital wall fractures, respectively. Bone defect area of the RP group was significantly reduced in comparison with that of the conventional group (8.03 ± 3.5% versus 18.7 ± 15.41% for medial wall fractures, 7.14 ± 5.74% versus 12.8 ± 4.92% for inferior wall fractures, respectively, p < 0.01). Satisfactory results were achieved regarding the layout angles and the gap lengths, presenting significantly reduced values in the RP group compared to that in the conventional group (p < 0.01). Conclusions: More accurate restoration of traumatic orbital wall fractures can be achieved using patientspecific 3D-printed RP skull models. © 2017 European Association for Cranio-Maxillo-Facial Surgery. Published by Elsevier Ltd. All rights reserved.
Keywords: 3D printing Blow out fracture Computer simulation
1. Introduction For traumatic orbital wall fractures, restoration of the orbital wall contour is challenging because of the complexity of the 3D anatomy of the orbit. Even with experienced surgeons, the manual reproduction of the contralateral unaffected orbit may result in clinical failures such as diplopia, enophthalmos and even blindness (Shin et al., 2013). With advances in 3D printing technologies and computer-aided design/manufacturing, precise preoperative planning for orbital reconstruction is now feasible. Computer-aided design software
* Corresponding author. Department of Plastic and Reconstructive Surgery, University of Ulsan College of Medicine, Asan Medical Center, 88, Olympic-ro 43-gil, SongPa-Gu, Seoul 05505, Republic of Korea. Fax: þ82 2 476 7471. E-mail address: [email protected] (T.S. Oh).
can be used for the reconstruction of 3D computed tomography images of the orbit. Such technologies allow the creation of a patient-specific implant. Recently these individualized 3D printed orbital implants have been used in maxillofacial surgery. Such implants, however, are difficult to fabricate, are expensive, and sometimes lack dimensional prevision (Stoor et al., 2014). Furthermore, they are difficult to adapt or sculpt during surgery because of their mechanical properties. In this study, we propose the combination of mirroring technique and rapid prototyping (RP) skull models using 3D printing, which offers an accurate and cost-effective treatment for orbital wall fractures. We hypothesize the use of RP skull modeling allows the precise molding and insetting of the implant compared with conventional free-hand techniques. The surgical outcomes were evaluated using computed tomography values which assess the accuracy of implant insertion and the degree of post-operative enophthalmos, diplopia and limitation of globe movement.
http://dx.doi.org/10.1016/j.jcms.2017.03.010 1010-5182/© 2017 European Association for Cranio-Maxillo-Facial Surgery. Published by Elsevier Ltd. All rights reserved.
Y.C. Kim et al. / Journal of Cranio-Maxillo-Facial Surgery 45 (2017) 928e936
2. Material and methods We retrospectively studied 82 patients who underwent insertion of a titanium mesh to repair an orbital wall fracture at the Department of Plastic Surgery of the Asan Medical Center in Seoul (South Korea) between 2012 and 2014. Inclusion criteria included; (1) unilateral isolated blow out fracture in medial or inferior orbit; (2) no previous history of orbital trauma; (3) surgery was performed within 1 month from occurrence of orbital fracture; (4) a follow up of at least 6 months. Patients with a zygomatico-maxillary fracture or nasal bone fracture were excluded. We performed a computed tomography (CT) scan of all patients preoperatively on their first visit and postoperatively on the second day. All studies on human material were approved by the ethics committees in our institution and were in agreement with the Declaration of Helsinki.
929
then an RP skull model (AMK, Inc., Goyang-si, Gyeonggi-do, Korea) representing the individual model of the uninjured state was made using a 3D printer (Projet 660 Pro; 3D Systems, Inc., Rock Hill, SC). The mirror-imaged RP skull model was used intraoperatively to provide the original contour of the orbit, onto which the titanium mesh was molded for the reconstruction of the fractured orbit (Fig. 1). Using preoperative CT scans, original and mirror-imaged RP skull models were made for 44 patients including 24 medial wall and 20 inferior wall fractures. Placement of the intraoperatively prebent porous polyethylene-coated titanium mesh (SynPOR; Synthes, Inc., West Chester, PA), molded on the 3D-printed RP skull template, was performed in the RP skull model group, referred to as RP group (Fig. 1). Whereas in the conventional group, the SynPOR implant was molded with a free-hand technique. Table 1 shows the demographic information for these groups. All surgical procedures were performed by a single surgeon (T.S.O.).
2.1. 3D simulation with mirroring techniques and RP skull modeling 2.2. Surgical procedures 3D simulation was performed using Mimics 3D software (Materialise NV, Inc., Leuven, Belgium). The normal unaffected orbit was reflected onto the contralateral side by mirroring techniques, and
The surgical approach was based on either transconjunctival or transcaruncular dissection according to the location of the orbital
Fig. 1. Titanium mesh intraoperatively pre-bent on a rapid prototyping model. A 3D image of the orbit was reproduced from a mirror image of the unaffected side and then a 3D printer was used to make a rapid prototyping model. The titanium mesh was intraoperatively molded on the 3D rapid prototyping model.
930
Y.C. Kim et al. / Journal of Cranio-Maxillo-Facial Surgery 45 (2017) 928e936
Table 1 Details of patient demographics in the RP group and conventional group.
N Age (years) Average Standard deviation Range Sex (M:F) Follow-up (months) Average Standard deviation Range Fracture location (n, %) Medial wall Inferior wall Cause of injury (n, %) Assaults Falls Sports related Vehicle accidents Time between trauma and surgery (days, mean ± SD)
RP group
Conventional group
44
38
38.7 15.8 13e71 3.3:1
34.9 12.6 15e61 3.1:1
15.9 5.7 6e27
16.1 6.7 6e30
24 20
22 16
17 15 8 4 11.2 ± 4.8
8 12 11 7 10.7 ± 5.1
wall fracture. The medial and inferior walls were fully exposed in a subperiosteal fashion and any herniated or entrapped orbital soft tissue was manually reduced. In all cases, a porous polyethylenecoated titanium mesh (SynPOR, Synthes, Inc., West Chester, PA) was used to cover the area of the bone defect. In the RP group, the implant was pre-bent on the mirror-imaged 3D-reconstructed RP skull model. Then, the prebent implant was placed on the area of the bone defect under direct visualization of the fracture segment using the original RP skull model. The implant was fixed with 4mm Matrix midface screws (Synthes, Inc., West Chester, PA) on
either the medial or inferior orbital rim according to the fracture type. In the conventional group, the above procedures were omitted and the implant was molded with a free-hand technique and placement was performed without any supportive tool. 2.3. Assessment of the accuracy of implant insertion Sung et al. reported that CT scans could predict and correlate the degree of enophthalmos and extent of a bone defect (2013). In the current study we performed CT scans and thereby assessed the accuracy of implant insertion by measuring the pre- and postoperative area of the defected bone (Fig. 2). In order to ensure that an implant correctly coincided with the original bony contour, we measured the layout angle, which was defined as an acute angle between the implant and an arbitrary line presenting the premorbid status of the orbital wall, to assess the accuracy of the insertion of an implant and its gap length, which was the distance from the distal tip of the implant and the posterior bony shelf margin (Figs. 3 and 4). Any changes of the area of the bone defect in pre- and post-operative state, the layout angle, the gap length and the operating time were compared between the RP group and conventional group. The ManneWhitney U test was used for statistical analysis. Clinical findings (enopthalmos, presence of diplopia, ocular motility) were assessed preoperatively and postoperatively at 3 month follow up visit. We assessed the globe position grossly from worm's eye view by drawing a horizontal line from mid-pupil of the unaffected eye to the affected eye and determining the vertical distance from mid-pupil to mid-pupil in mm. Enophthalmos was defined as the posterior displacement of the affected eye relative to the unaffected side by measuring the difference in each side. The evaluation scale consisted of four grades from 0 to 3 as follows:
Fig. 2. Measurement of the preoperative and postoperative orbital wall defect areas. We defined the length of the orbital wall defect as ‘a’ and the area of the bone defect in each P section by multiplying width and height (an 1.0 mm). We then used the formula A ¼ N n¼1 an to sum the areas of the bone defects for each section. Using this method, we measured the preoperative and postoperative orbital wall defect areas (Lee et al., 2014). (a, c) Preoperative and postoperative orbital wall defect (am 1.0 mm) was measured in patients with medial wall fracture. (b, d) Preoperative and postoperative orbital wall defect (ai 1.0 mm) was measured in patients with inferior wall fracture.
Y.C. Kim et al. / Journal of Cranio-Maxillo-Facial Surgery 45 (2017) 928e936
931
Fig. 3. Measurement of the Layout angle of the titanium mesh. To measure the layout angle of the titanium mesh, we extended an arbitrary line from the starting point of the medial and inferior orbital wall fracture to its end. This line (blank arrow) reflects the normal bony contour and is easy to suppose from contralateral normal orbit (arrow). (a, b) We used sagittal CT images for inferior wall fractures and (c) coronal CT images for medial wall fractures. The maximum angle between the inserted mesh and the arbitrary line, defined as ai for inferior wall fracture and am for medial wall fracture, was obtained in postoperative computed tomography scans.
grade 0, no enophthalmos; grade 1, mild enophthalmos (ranged less than 1 mm); grade 2, moderate enophthalmos (ranged from 1 to 2 mm); grade 3, severe enophthalmos (ranged over 3 mm). Presence of diplopia was assessed in all gaze positions. Ocular motility was assessed by four positions of gaze (adduction, abduction, supraduction, and infraduction) and was graded on a scale from 0 to 4 (0, no limitation; 1, duction of 30 e45 ; 2, duction of 15 e30 ; 3, duction of <15 ; 4, no movement). 2.4. Measurement of the bone defect area at the medial and inferior orbital wall fracture sites CT scans were performed preoperatively and on day two of postoperation with a slice thickness of 1.0 mm to analyze the extent of the defect of the orbital wall. Using a picture archiving and communication system, the defect areas of the medial wall and inferior wall were measured in coronal and sagittal sections, respectively. As shown in Fig. 2, we measured the length of the line starting from the orbital wall fracture to its end for medial and inferior wall fractures and designated this parameter ‘a’. We multiplied ‘a’ by the CT scan interval (1.0 mm) to obtain the area of the bone defect. By summing the measurements of each area, the total area of the bone defect could be measured and was designated ‘A’ (Lee et al., 2014). 2.5. Measurement of the layout angle of titanium mesh at the medial and inferior orbital wall fracture site The layout angle of the titanium mesh was determined by the acute angle between the inserted mesh and an arbitrary line
connecting the starting point of the orbital wall fracture to its end (Fig. 3). This line reflects the normal bony contour and is easy to suppose from contralateral normal orbit. We used coronal CT images for medial wall fractures and sagittal CT images for inferior wall fractures. The maximum angle was obtained in postoperative computed tomography scans. The layout angle was compared between the RP group and conventional groups. 2.6. Measurement of the gap length between the mesh and the bony margin of medial and inferior orbital wall fractures To measure the gap length between the inserted mesh and the fracture element, the bony margin of the fracture element was defined as the anterior and posterior part. The length between the mesh and the orbital wall fracture was defined as the distance from the tip of the inserted mesh to the posterior bony margin (Fig. 4). The maximum length was obtained in postoperative computed tomography scans. The gap lengths were compared between the RP group and conventional group. 3. Results The preoperative and postoperative orbital wall defect areas were measured in the two groups. For medial wall fractures, the mean values of the preoperative and postoperative orbital wall defect areas were 1.64 cm2 (ranged from 0.96 to 2.66 cm2) and 0.13 cm2 (ranged from 0 to 0.28 cm2) in the RP group and 1.87 cm2 (ranged from 1.05 to 2.84 cm2) and 0.42 cm2 (ranged from 0 to 2.1 cm2) in the conventional group, respectively (Table 2). A significant difference between the preoperative and postoperative
932
Y.C. Kim et al. / Journal of Cranio-Maxillo-Facial Surgery 45 (2017) 928e936
Fig. 4. Measurement of the Gap length between the inserted mesh and the bony margin of orbital wall fractures. The Gap length was defined as the maximum distance from the tip of the inserted mesh to the posterior bony margin and presents the insufficient length covering the posterior defect. (a, b) Each value, described as ‘L’, was obtained in coronal sections for medial wall fractures (Lm) and in sagittal sections for inferior wall fractures (Li).
bone defects between the two groups was observed, where a mean value of 8.03% in the RP skull model group was obtained compared with 18.7% in the conventional group (p < 0.01, Fig. 5). For inferior wall fractures, the mean values of the preoperative and postoperative orbital wall defect areas were 1.79 cm2 (ranged from 1.01 to 2.78 cm2) and 0.12 cm2 (ranged from 0 to 0.4 cm2) in the RP group and 1.92 cm2 (ranged from 1.03 to 2.96 cm2) and 0.41 cm2 (ranged from 0 to 1.19 cm2) in the conventional group, respectively (Table 3). Also, a significant difference of the defect area changes between the preoperative and postoperative defect areas were identified with mean values of 7.14% in the RP skull model group and 12.8% in the conventional group (p < 0.01, Fig. 6). We measured the layout angle of the titanium mesh at the medial and inferior orbital wall fracture site. For medial wall fractures, the average layout angle of the inserted mesh was 3.49 (ranged from 0.1 to 6.84 ) in the RP group and 9.03 (ranged from 1.47 to 21.76 ) in the conventional group. For inferior wall fractures, the average layout angle of the inserted mesh was 2.23 in the RP skull model group and 4.69 in the conventional group. For both types of fractures, the layout angles showed a significant difference between the RP and the conventional groups (Table 3). The gap length was measured between the inserted mesh and the fracture element. For medial wall fractures, the average lengths were 3.1 mm (ranged from 1.47 to 21.76 mm) in the RP group and 6.37 mm (ranged from 1.47 to 21.76 ) in the conventional group, which showed a significant difference between the two
groups (p < 0.01). For inferior wall fractures, the average length was 5.21 mm (ranged from 1.47 to 21.76 ) in the RP skull model group and 6.64 mm (ranged from 1.47 to 21.76 ) in the conventional group. No significant differences were noted between the two groups in inferior wall fracture type (Table 3). The operating time was defined from the point of skin incision to the end of skin closure. It was measured in all patients, and the mean values were compared between the two groups. The operating times were 75.6 min for the RP group and 77.7 min for the conventional group. No significant differences were identified between the two groups in the duration of surgery (p ¼ 0.519). 44 patients in the RP group and 38 patients in the conventional group were available at the 3-month follow-up visit to the clinic. Two patients in the RP group had mild enophthalmos, which partially improved over time. Whereas in the conventional group, 3 patients had mild enophthalmos and 2 patients had moderate and severe enophthalmos, respectively. One patient in the RP group and 4 patients in the conventional group respectively had mild ocular limitation, which gradually resolved. Two patients in the conventional group who had moderate to severe enophthalmos and limited ocular motility underwent revisional operation at 6 months postoperatively. In these patients, the malpositioned implant did not fully cover the bone defect and the entrapped soft tissue component was reduced. Two patients in the RP group and 5 patients in the conventional group had diplopia in some position of the gaze postoperatively. Of these, 2 patients in the RP group and 3
Y.C. Kim et al. / Journal of Cranio-Maxillo-Facial Surgery 45 (2017) 928e936
933
Table 2 Measurements of the preoperative and postoperative bone defect area for medial and inferior wall fractures. Bone defect area (cm2)
Medial wall fracture RP group Conventional group Inferior wall fracture RP group Conventional group a
n
Preoperative (Mean ± SD)
Postoperative (Mean ± SD)
Defect area changes (D, Mean ± SD, %)
p value
24 22
1.64 ± 0.42 1.87 ± 0.66
0.13 ± 0.05 0.42 ± 0.54
8.03 ± 3.5 18.7 ± 15.41
p < 0.01a
20 16
1.79 ± 0.59 1.92 ± 0.59
0.12 ± 0.11 0.41 ± 0.4
7.14 ± 5.74 12.8 ± 4.92
p < 0.01a
Significant difference of Defect area changes was identified between the two groups.
Fig. 5. Comparison of Bone defect area between the RP and conventional groups for medial wall fractures. y No statistical significance of the preoperative bone defect area between the RP and the conventional group (p ¼ 0.167). * Significant difference of postoperative bone defect area between the RP and conventional group is identified (p ¼ 0.02). Defect area change (%) was 18.7% in the conventional group and 8.03% in the RP group, and these values showed significant difference (p < 0.01).
patients in the conventional group had complete resolution at 6 month follow up, and the other 2 patients in the conventional group had residual diplopia that required revisional operation (Table 4). 4. Discussion The challenges associated with the management of orbital wall fractures originate from the complex skeletal and soft tissue anatomy of the orbit. It is important to consider the unique contours of the orbital anatomy, and this anatomical region must be precisely corrected to reconstitute the orbital volume and contour (Gordon et al., 2012). Incorrect reconstruction of the orbital dimensions, whether under- or over-corrected, may cause severe posttraumatic orbital deformities such as enophthalmos, diplopia, and
visual acuity disturbance (Manolidis et al., 2002; Shin et al., 2013). For reconstruction of medial wall and orbital floor fractures, the implants should be prepared to fit this exact contour, but this goal is difficult to achieve by free-hand implant bending. Given that the implant may need to be inserted and removed two or three times before the best shape and orientation are determined, this can place the periorbital soft tissues at increased risk of iatrogenic injury. Without additional tools such as intraoperative imaging or navigation, the placement of the implants is likely to be incorrect (Schon et al., 2006; Stoetzer et al., 2011). The posterior orbital floor and medial orbital wall are the key areas of the orbit with their own unique contours. The contour of the orbital floor slopes down directly behind the infraorbital rim, before ascending toward the posterior aspect, resulting in an Sshaped orbital floor in the posterior third of the orbit. When
Table 3 Comparisons of postoperative layout angle and gap length for medial and inferior wall fractures.
Medial wall fracture RP group conventional group Inferior wall fracture RP group conventional group a b
n
Layout angle (degree, Mean ± SD)
p value
Gap length (mm, Mean ± SD)
p value
24 22
3.49 ± 1.97 9.03 ± 4.9
p < 0.01a
3.1 ± 1.46 6.37 ± 3.29
p < 0.01b
20 16
2.23 ± 1.37 4.69 ± 2.51
p < 0.01a
5.21 ± 2.67 6.64 ± 3.61
p ¼ 0.12
Significant difference of layout angle was identified between the two groups. Significant difference of gap length was identified between the two groups.
934
Y.C. Kim et al. / Journal of Cranio-Maxillo-Facial Surgery 45 (2017) 928e936
Fig. 6. Comparison of Bone defect area between the RP and conventional groups for inferior wall fractures. y No statistical significance of the preoperative bone defect area between the RP and the conventional group (p ¼ 0.54). * Significant difference of postoperative bone defect area between the RP and conventional group is identified (p ¼ 0.01). Defect area change (%) was 12.8% in the conventional group and 7.14% in the RP group and these values showed significant difference (p < 0.01).
moving from the orbital floor to the medial orbital wall, the orbital floor inclines superiorly to meet the medial wall, creating a bony convergence in the posteromedial orbit. Restoration of this unique shape of the orbit following traumatic injury may often be difficult due to the inability of the surgeon to obtain adequate visibility and to verify proper implant placement during the operation (Ozer et al., 2016). One of the greatest advantages of using the RP skull model is that it permits direct visualization of the orbit, allowing surgeons to customize the patient-specific implants and simulate the placement of the implant onto the premorbid model. Most surgeons hesitate to dissect the posterior space of the orbit, in which optic nerve and ophthalmic vessels can be damaged. With the guide of RP skull model, surgeons are able to predict how far the implant can be inserted, where the exact location of the defect is, thereby achieving sufficient coverage of the defect even at the posterior side of the orbit (Fig. 7). For precise results, the orbital space can be volumetrically assessed using computer software. We described, in our previous reports of 127 traumatic orbital reconstructions, that precise volumetric reconstruction could be achieved using the mirrorimaged RP skull model (Park et al., 2015). Considering the clinical conditions, however, it is hard to access such software. 2-
Table 4 Clinical outcomes at postoperative 3 month visit in each group. Sign/symptom
RP group (n ¼ 44)
Enophthalmos None 95.45% (42/44) Mild 4.54% (2/44) Moderate 0 Severe 0 Presence of diplopia 4.54% (2/44) Limitation of ocular motility 0 97.72% (43/44) 1 2.27% (1/44) 2 0 3 0 4 0
Conventional group (n ¼ 38) 86.84% (33/38) 7.89% (3/38) 2.63% (1/38) 2.63% (1/38) 13.15% (5/38) 84.21% (32/38) 10.52% (4/38) 5.26% (2/38) 0 0
dimensional CT scans can also provide reliable information on the extent of the fracture and thereby can be used more practically for clinical situations (Ploder et al., 2002). We used three parameters to compare the surgical outcomes, and the overall achievement was much improved with the use of the RP skull models. The computed tomography parameters in our present study were easy to use and had some clinical value. The degree of enophthalmos could be predicted based on the area of the bone defect. Sung et al. (2013) suggested that an area of a bone defect of approximately 2.75 cm2 is likely to correspond to 2 mm of enophthalmos. In our current study, although the area of the bone defect was corrected properly in both groups, the mean reduction in the area of the bone defect was significantly better in the RP skull model group. This is most likely because the implant was molded as closely as possible to the original shape by direct visualization of the posterior bony shelf considering its unique bulging contour where it would have been limited to access it normally by conventional methods. It was possible to correlate the layout angle to the degree of enophthalmos in incorrectly placed implants as, even with a minimal change, it could cause a relatively large volumetric discrepancy in the original orbit. The orbital volumes after an implant insertion are likely to be smaller than the premorbid volumes. This can be explained by internal orbital placement of the implant and because the added thickness of the mesh result in an overall reduction in the orbital volume (Strong et al., 2013). Thus, the correct axis of the implant placement can be guaranteed by repetitive trials using the RP skull models before an implant insertion. Further analysis of the relationship between the layout angle and enophthalmos might suggest the ideal axis of implant position that would not increase enophthalmos. More recent developments in computer-aided design/ manufacturing have paved the way for patient-specific implants that provide more accurate restoration of the internal orbit and minimize the need for major intraoperative manipulations (Metzger et al., 2006). Although patient-specific implants seem to offer the best reconstructive contour, their fabrication is expensive and time-consuming (Strong et al., 2013). To date, 3D printed titanium orbital implants can cost between $ 3,000 and $ 5,000 (Vehmeijer et al., 2016). Otherwise, the costs of RP skull models
Y.C. Kim et al. / Journal of Cranio-Maxillo-Facial Surgery 45 (2017) 928e936
935
Fig. 7. Supportive role of using RP skull model when covering the defect at the posterior side of orbit. (a, b) Preoperative CT scans for medial wall and inferior wall fracture, respectively. (c, d) Postoperative CT scans showed the bone defect in the posterior orbit (arrows) can be sufficiently covered with the implants under guidance of RP skull model.
ranged from $ 300 and $ 500. Free-hand bent implants are more cost-effective but have lower contour accuracy and ease of use (Strong et al., 2013). There is no international consensus on the ideal material to be used for orbital reconstruction, however, the success of orbital fracture repair depends not only on the use of material that restores the premorbid orbital volume and shape, but also the surgeon's capacity to visualize the geometry of the bony defect spatially and to place the implants to fit the original contour. We felt that 3D-printed RP skull models could supply the geometric information of the fractured orbit and allowed the anatomical placement of implant materials until they fitted the defect as precisely as possible. The RP skull model is influenced by many factors affecting the stereolithographic process, therefore artifacts can occur because of poor equipment, patient movement, or the presence of metallic implants. Careful patient positioning and radiographic techniques can minimize these problems (Chang et al., 2003). Given the occurrence of artifacts that fail to reflect the thin orbital wall, especially through the lamina papyracea, CT slices less than 1.5 mm in thickness can minimize these errors. We hypothesized that the use of RP skull models would reduce the time of surgery by eliminating time spent on molding and repetitive fitting. However, there was no significant difference in operation time between the RP and conventional groups. 5. Conclusion More accurate restoration of traumatic orbital wall fractures can be achieved using patient-specific 3D-printed RP skull models.
Surgical outcomes assessed using computed tomography parameters show better results for RP skull model groups than models molded with a free-hand technique. Conflict of interest The authors have no financial interest in any of the products or devices mentioned in this article. Acknowledgments All persons who have contributed to the study are listed as authors, since everyone has met the listed criteria for authorship. There exist no current funding sources for this study. References Chang PS, Parker TH, Patrick Jr CW, Miller MJ: The accuracy of stereolithography in planning craniofacial bone replacement. J Craniofac Surg 14: 164e170, 2003 Gordon CR, Susarla SM, Yaremchuk MJ: Quantitative assessment of medial orbit fracture repair using computer-designed anatomical plates. Plast Reconstr Surg 130: 698ee705e, 2012 Lee KM, Park JU, Kwon ST, Kim SW, Jeong EC: Three-dimensional pre-bent titanium implant for concomitant orbital floor and medial wall fractures in an East asian population. Arch Plast Surg 41: 480e485, 2014 Manolidis S, Weeks BH, Kirby M, Scarlett M, Hollier L: Classification and surgical management of orbital fractures: experience with 111 orbital reconstructions. J Craniofac Surg 13: 726e738, 2002 Metzger MC, Schon R, Weyer N, Rafii A, Gellrich NC, Schmelzeisen R, et al: Anatomical 3-dimensional pre-bent titanium implant for orbital floor fractures. Ophthalmology 113: 1863e1868, 2006 Ozer MA, Govsa F, Kazak Z, Erdogmus S, Celik S: Redesign and treatment planning orbital floor reconstruction using computer analysis anatomical landmarks. Eur Arch Otorhinolaryngol 273: 2185e2191, 2016
936
Y.C. Kim et al. / Journal of Cranio-Maxillo-Facial Surgery 45 (2017) 928e936
Park SW, Choi JW, Koh KS, Oh TS: Mirror-imaged rapid prototype skull model and pre-molded synthetic scaffold to achieve optimal orbital cavity reconstruction. J Oral Maxillofac Surg 73: 1540e1553, 2015 Ploder O, Klug C, Backfrieder W, Voracek M, Czerny C, Tschabitscher M: 2D- and 3Dbased measurements of orbital floor fractures from CT scans. J Craniomaxillofac Surg 30: 153e159, 2002 Schon R, Metzger MC, Zizelmann C, Weyer N, Schmelzeisen R: Individually preformed titanium mesh implants for a true-to-original repair of orbital fractures. Int J Oral Maxillofac Surg 35: 990e995, 2006 Shin JW, Lim JS, Yoo G, Byeon JH: An analysis of pure blowout fractures and associated ocular symptoms. J Craniofac Surg 24: 703e707, 2013 Stoetzer M, Rana M, von See C, Eckardt AM, Gellrich NC: Reconstruction of defects of maxillary sinus wall after removal of a huge odontogenic lesion using prebended 3D titanium-mesh and CAD/CAM technique. Head Face Med 7: 21, 2011
Stoor P, Suomalainen A, Lindqvist C, Mesimaki K, Danielsson D, Westermark A, et al: Rapid prototyped patient specific implants for reconstruction of orbital wall defects. J Craniomaxillofac Surg 42: 1644e1649, 2014 Strong EB, Fuller SC, Wiley DF, Zumbansen J, Wilson MD, Metzger MC: Preformed vs intraoperative bending of titanium mesh for orbital reconstruction. Otolaryngol Head Neck Surg 149: 60e66, 2013 Sung YS, Chung CM, Hong IP: The correlation between the degree of enophthalmos and the extent of fracture in medial orbital wall fracture left untreated for over six months: a retrospective analysis of 81 cases at a single institution. Arch Plast Surg 40: 335e340, 2013 Vehmeijer M, van Eijnatten M, Liberton N, Wolff J: A novel method of orbital floor reconstruction using virtual planning, 3-dimensional printing, and autologous bone. J Oral Maxillofac Surg 74: 1608e1612, 2016
Contents lists available at ScienceDirect
Journal of Cranio-Maxillo-Facial Surgery journal homepage: www.jcmfs.com
The accuracy of patient specific implant prebented with 3D-printed rapid prototype model for orbital wall reconstruction Young Chul Kim a, Woo Shik Jeong a, Tae-kyung Park b, Jong Woo Choi a, Kyung S. Koh a, Tae Suk Oh a, * a b
Department of Plastic Surgery, Asan Medical Center, University of Ulsan College of Medicine, Seoul, Republic of Korea Department of Emergency, The Newcastle Upon Tyne Hospitals, Newcastle Upon Tyne, United Kingdom
a r t i c l e i n f o
a b s t r a c t
Article history: Paper received 22 August 2016 Accepted 20 March 2017 Available online 27 March 2017
Background: This study evaluated the accuracy of blow out fracture reduction using 3D-printed rapid prototyping (RP) skull modeling. Patients and methods: Retrospective review was performed for 82 patients who underwent posttraumatic orbital wall fracture reduction between 2012 and 2014. Patients were divided into two groups according to the use of 3D-printed RP skull model reproduced by mirroring technique, onto which a titanium mesh was anatomically molded. Using computed tomographic scans, the areas of preand post-operative orbital wall defect, the layout angles of the titanium mesh, and the gap lengths between the implant and fracture margin were compared between the two groups. Results: Of the 82 patients identified, 46 and 36 were diagnosed with medial and inferior orbital wall fractures, respectively. Bone defect area of the RP group was significantly reduced in comparison with that of the conventional group (8.03 ± 3.5% versus 18.7 ± 15.41% for medial wall fractures, 7.14 ± 5.74% versus 12.8 ± 4.92% for inferior wall fractures, respectively, p < 0.01). Satisfactory results were achieved regarding the layout angles and the gap lengths, presenting significantly reduced values in the RP group compared to that in the conventional group (p < 0.01). Conclusions: More accurate restoration of traumatic orbital wall fractures can be achieved using patientspecific 3D-printed RP skull models. © 2017 European Association for Cranio-Maxillo-Facial Surgery. Published by Elsevier Ltd. All rights reserved.
Keywords: 3D printing Blow out fracture Computer simulation
1. Introduction For traumatic orbital wall fractures, restoration of the orbital wall contour is challenging because of the complexity of the 3D anatomy of the orbit. Even with experienced surgeons, the manual reproduction of the contralateral unaffected orbit may result in clinical failures such as diplopia, enophthalmos and even blindness (Shin et al., 2013). With advances in 3D printing technologies and computer-aided design/manufacturing, precise preoperative planning for orbital reconstruction is now feasible. Computer-aided design software
* Corresponding author. Department of Plastic and Reconstructive Surgery, University of Ulsan College of Medicine, Asan Medical Center, 88, Olympic-ro 43-gil, SongPa-Gu, Seoul 05505, Republic of Korea. Fax: þ82 2 476 7471. E-mail address: [email protected] (T.S. Oh).
can be used for the reconstruction of 3D computed tomography images of the orbit. Such technologies allow the creation of a patient-specific implant. Recently these individualized 3D printed orbital implants have been used in maxillofacial surgery. Such implants, however, are difficult to fabricate, are expensive, and sometimes lack dimensional prevision (Stoor et al., 2014). Furthermore, they are difficult to adapt or sculpt during surgery because of their mechanical properties. In this study, we propose the combination of mirroring technique and rapid prototyping (RP) skull models using 3D printing, which offers an accurate and cost-effective treatment for orbital wall fractures. We hypothesize the use of RP skull modeling allows the precise molding and insetting of the implant compared with conventional free-hand techniques. The surgical outcomes were evaluated using computed tomography values which assess the accuracy of implant insertion and the degree of post-operative enophthalmos, diplopia and limitation of globe movement.
http://dx.doi.org/10.1016/j.jcms.2017.03.010 1010-5182/© 2017 European Association for Cranio-Maxillo-Facial Surgery. Published by Elsevier Ltd. All rights reserved.
Y.C. Kim et al. / Journal of Cranio-Maxillo-Facial Surgery 45 (2017) 928e936
2. Material and methods We retrospectively studied 82 patients who underwent insertion of a titanium mesh to repair an orbital wall fracture at the Department of Plastic Surgery of the Asan Medical Center in Seoul (South Korea) between 2012 and 2014. Inclusion criteria included; (1) unilateral isolated blow out fracture in medial or inferior orbit; (2) no previous history of orbital trauma; (3) surgery was performed within 1 month from occurrence of orbital fracture; (4) a follow up of at least 6 months. Patients with a zygomatico-maxillary fracture or nasal bone fracture were excluded. We performed a computed tomography (CT) scan of all patients preoperatively on their first visit and postoperatively on the second day. All studies on human material were approved by the ethics committees in our institution and were in agreement with the Declaration of Helsinki.
929
then an RP skull model (AMK, Inc., Goyang-si, Gyeonggi-do, Korea) representing the individual model of the uninjured state was made using a 3D printer (Projet 660 Pro; 3D Systems, Inc., Rock Hill, SC). The mirror-imaged RP skull model was used intraoperatively to provide the original contour of the orbit, onto which the titanium mesh was molded for the reconstruction of the fractured orbit (Fig. 1). Using preoperative CT scans, original and mirror-imaged RP skull models were made for 44 patients including 24 medial wall and 20 inferior wall fractures. Placement of the intraoperatively prebent porous polyethylene-coated titanium mesh (SynPOR; Synthes, Inc., West Chester, PA), molded on the 3D-printed RP skull template, was performed in the RP skull model group, referred to as RP group (Fig. 1). Whereas in the conventional group, the SynPOR implant was molded with a free-hand technique. Table 1 shows the demographic information for these groups. All surgical procedures were performed by a single surgeon (T.S.O.).
2.1. 3D simulation with mirroring techniques and RP skull modeling 2.2. Surgical procedures 3D simulation was performed using Mimics 3D software (Materialise NV, Inc., Leuven, Belgium). The normal unaffected orbit was reflected onto the contralateral side by mirroring techniques, and
The surgical approach was based on either transconjunctival or transcaruncular dissection according to the location of the orbital
Fig. 1. Titanium mesh intraoperatively pre-bent on a rapid prototyping model. A 3D image of the orbit was reproduced from a mirror image of the unaffected side and then a 3D printer was used to make a rapid prototyping model. The titanium mesh was intraoperatively molded on the 3D rapid prototyping model.
930
Y.C. Kim et al. / Journal of Cranio-Maxillo-Facial Surgery 45 (2017) 928e936
Table 1 Details of patient demographics in the RP group and conventional group.
N Age (years) Average Standard deviation Range Sex (M:F) Follow-up (months) Average Standard deviation Range Fracture location (n, %) Medial wall Inferior wall Cause of injury (n, %) Assaults Falls Sports related Vehicle accidents Time between trauma and surgery (days, mean ± SD)
RP group
Conventional group
44
38
38.7 15.8 13e71 3.3:1
34.9 12.6 15e61 3.1:1
15.9 5.7 6e27
16.1 6.7 6e30
24 20
22 16
17 15 8 4 11.2 ± 4.8
8 12 11 7 10.7 ± 5.1
wall fracture. The medial and inferior walls were fully exposed in a subperiosteal fashion and any herniated or entrapped orbital soft tissue was manually reduced. In all cases, a porous polyethylenecoated titanium mesh (SynPOR, Synthes, Inc., West Chester, PA) was used to cover the area of the bone defect. In the RP group, the implant was pre-bent on the mirror-imaged 3D-reconstructed RP skull model. Then, the prebent implant was placed on the area of the bone defect under direct visualization of the fracture segment using the original RP skull model. The implant was fixed with 4mm Matrix midface screws (Synthes, Inc., West Chester, PA) on
either the medial or inferior orbital rim according to the fracture type. In the conventional group, the above procedures were omitted and the implant was molded with a free-hand technique and placement was performed without any supportive tool. 2.3. Assessment of the accuracy of implant insertion Sung et al. reported that CT scans could predict and correlate the degree of enophthalmos and extent of a bone defect (2013). In the current study we performed CT scans and thereby assessed the accuracy of implant insertion by measuring the pre- and postoperative area of the defected bone (Fig. 2). In order to ensure that an implant correctly coincided with the original bony contour, we measured the layout angle, which was defined as an acute angle between the implant and an arbitrary line presenting the premorbid status of the orbital wall, to assess the accuracy of the insertion of an implant and its gap length, which was the distance from the distal tip of the implant and the posterior bony shelf margin (Figs. 3 and 4). Any changes of the area of the bone defect in pre- and post-operative state, the layout angle, the gap length and the operating time were compared between the RP group and conventional group. The ManneWhitney U test was used for statistical analysis. Clinical findings (enopthalmos, presence of diplopia, ocular motility) were assessed preoperatively and postoperatively at 3 month follow up visit. We assessed the globe position grossly from worm's eye view by drawing a horizontal line from mid-pupil of the unaffected eye to the affected eye and determining the vertical distance from mid-pupil to mid-pupil in mm. Enophthalmos was defined as the posterior displacement of the affected eye relative to the unaffected side by measuring the difference in each side. The evaluation scale consisted of four grades from 0 to 3 as follows:
Fig. 2. Measurement of the preoperative and postoperative orbital wall defect areas. We defined the length of the orbital wall defect as ‘a’ and the area of the bone defect in each P section by multiplying width and height (an 1.0 mm). We then used the formula A ¼ N n¼1 an to sum the areas of the bone defects for each section. Using this method, we measured the preoperative and postoperative orbital wall defect areas (Lee et al., 2014). (a, c) Preoperative and postoperative orbital wall defect (am 1.0 mm) was measured in patients with medial wall fracture. (b, d) Preoperative and postoperative orbital wall defect (ai 1.0 mm) was measured in patients with inferior wall fracture.
Y.C. Kim et al. / Journal of Cranio-Maxillo-Facial Surgery 45 (2017) 928e936
931
Fig. 3. Measurement of the Layout angle of the titanium mesh. To measure the layout angle of the titanium mesh, we extended an arbitrary line from the starting point of the medial and inferior orbital wall fracture to its end. This line (blank arrow) reflects the normal bony contour and is easy to suppose from contralateral normal orbit (arrow). (a, b) We used sagittal CT images for inferior wall fractures and (c) coronal CT images for medial wall fractures. The maximum angle between the inserted mesh and the arbitrary line, defined as ai for inferior wall fracture and am for medial wall fracture, was obtained in postoperative computed tomography scans.
grade 0, no enophthalmos; grade 1, mild enophthalmos (ranged less than 1 mm); grade 2, moderate enophthalmos (ranged from 1 to 2 mm); grade 3, severe enophthalmos (ranged over 3 mm). Presence of diplopia was assessed in all gaze positions. Ocular motility was assessed by four positions of gaze (adduction, abduction, supraduction, and infraduction) and was graded on a scale from 0 to 4 (0, no limitation; 1, duction of 30 e45 ; 2, duction of 15 e30 ; 3, duction of <15 ; 4, no movement). 2.4. Measurement of the bone defect area at the medial and inferior orbital wall fracture sites CT scans were performed preoperatively and on day two of postoperation with a slice thickness of 1.0 mm to analyze the extent of the defect of the orbital wall. Using a picture archiving and communication system, the defect areas of the medial wall and inferior wall were measured in coronal and sagittal sections, respectively. As shown in Fig. 2, we measured the length of the line starting from the orbital wall fracture to its end for medial and inferior wall fractures and designated this parameter ‘a’. We multiplied ‘a’ by the CT scan interval (1.0 mm) to obtain the area of the bone defect. By summing the measurements of each area, the total area of the bone defect could be measured and was designated ‘A’ (Lee et al., 2014). 2.5. Measurement of the layout angle of titanium mesh at the medial and inferior orbital wall fracture site The layout angle of the titanium mesh was determined by the acute angle between the inserted mesh and an arbitrary line
connecting the starting point of the orbital wall fracture to its end (Fig. 3). This line reflects the normal bony contour and is easy to suppose from contralateral normal orbit. We used coronal CT images for medial wall fractures and sagittal CT images for inferior wall fractures. The maximum angle was obtained in postoperative computed tomography scans. The layout angle was compared between the RP group and conventional groups. 2.6. Measurement of the gap length between the mesh and the bony margin of medial and inferior orbital wall fractures To measure the gap length between the inserted mesh and the fracture element, the bony margin of the fracture element was defined as the anterior and posterior part. The length between the mesh and the orbital wall fracture was defined as the distance from the tip of the inserted mesh to the posterior bony margin (Fig. 4). The maximum length was obtained in postoperative computed tomography scans. The gap lengths were compared between the RP group and conventional group. 3. Results The preoperative and postoperative orbital wall defect areas were measured in the two groups. For medial wall fractures, the mean values of the preoperative and postoperative orbital wall defect areas were 1.64 cm2 (ranged from 0.96 to 2.66 cm2) and 0.13 cm2 (ranged from 0 to 0.28 cm2) in the RP group and 1.87 cm2 (ranged from 1.05 to 2.84 cm2) and 0.42 cm2 (ranged from 0 to 2.1 cm2) in the conventional group, respectively (Table 2). A significant difference between the preoperative and postoperative
932
Y.C. Kim et al. / Journal of Cranio-Maxillo-Facial Surgery 45 (2017) 928e936
Fig. 4. Measurement of the Gap length between the inserted mesh and the bony margin of orbital wall fractures. The Gap length was defined as the maximum distance from the tip of the inserted mesh to the posterior bony margin and presents the insufficient length covering the posterior defect. (a, b) Each value, described as ‘L’, was obtained in coronal sections for medial wall fractures (Lm) and in sagittal sections for inferior wall fractures (Li).
bone defects between the two groups was observed, where a mean value of 8.03% in the RP skull model group was obtained compared with 18.7% in the conventional group (p < 0.01, Fig. 5). For inferior wall fractures, the mean values of the preoperative and postoperative orbital wall defect areas were 1.79 cm2 (ranged from 1.01 to 2.78 cm2) and 0.12 cm2 (ranged from 0 to 0.4 cm2) in the RP group and 1.92 cm2 (ranged from 1.03 to 2.96 cm2) and 0.41 cm2 (ranged from 0 to 1.19 cm2) in the conventional group, respectively (Table 3). Also, a significant difference of the defect area changes between the preoperative and postoperative defect areas were identified with mean values of 7.14% in the RP skull model group and 12.8% in the conventional group (p < 0.01, Fig. 6). We measured the layout angle of the titanium mesh at the medial and inferior orbital wall fracture site. For medial wall fractures, the average layout angle of the inserted mesh was 3.49 (ranged from 0.1 to 6.84 ) in the RP group and 9.03 (ranged from 1.47 to 21.76 ) in the conventional group. For inferior wall fractures, the average layout angle of the inserted mesh was 2.23 in the RP skull model group and 4.69 in the conventional group. For both types of fractures, the layout angles showed a significant difference between the RP and the conventional groups (Table 3). The gap length was measured between the inserted mesh and the fracture element. For medial wall fractures, the average lengths were 3.1 mm (ranged from 1.47 to 21.76 mm) in the RP group and 6.37 mm (ranged from 1.47 to 21.76 ) in the conventional group, which showed a significant difference between the two
groups (p < 0.01). For inferior wall fractures, the average length was 5.21 mm (ranged from 1.47 to 21.76 ) in the RP skull model group and 6.64 mm (ranged from 1.47 to 21.76 ) in the conventional group. No significant differences were noted between the two groups in inferior wall fracture type (Table 3). The operating time was defined from the point of skin incision to the end of skin closure. It was measured in all patients, and the mean values were compared between the two groups. The operating times were 75.6 min for the RP group and 77.7 min for the conventional group. No significant differences were identified between the two groups in the duration of surgery (p ¼ 0.519). 44 patients in the RP group and 38 patients in the conventional group were available at the 3-month follow-up visit to the clinic. Two patients in the RP group had mild enophthalmos, which partially improved over time. Whereas in the conventional group, 3 patients had mild enophthalmos and 2 patients had moderate and severe enophthalmos, respectively. One patient in the RP group and 4 patients in the conventional group respectively had mild ocular limitation, which gradually resolved. Two patients in the conventional group who had moderate to severe enophthalmos and limited ocular motility underwent revisional operation at 6 months postoperatively. In these patients, the malpositioned implant did not fully cover the bone defect and the entrapped soft tissue component was reduced. Two patients in the RP group and 5 patients in the conventional group had diplopia in some position of the gaze postoperatively. Of these, 2 patients in the RP group and 3
Y.C. Kim et al. / Journal of Cranio-Maxillo-Facial Surgery 45 (2017) 928e936
933
Table 2 Measurements of the preoperative and postoperative bone defect area for medial and inferior wall fractures. Bone defect area (cm2)
Medial wall fracture RP group Conventional group Inferior wall fracture RP group Conventional group a
n
Preoperative (Mean ± SD)
Postoperative (Mean ± SD)
Defect area changes (D, Mean ± SD, %)
p value
24 22
1.64 ± 0.42 1.87 ± 0.66
0.13 ± 0.05 0.42 ± 0.54
8.03 ± 3.5 18.7 ± 15.41
p < 0.01a
20 16
1.79 ± 0.59 1.92 ± 0.59
0.12 ± 0.11 0.41 ± 0.4
7.14 ± 5.74 12.8 ± 4.92
p < 0.01a
Significant difference of Defect area changes was identified between the two groups.
Fig. 5. Comparison of Bone defect area between the RP and conventional groups for medial wall fractures. y No statistical significance of the preoperative bone defect area between the RP and the conventional group (p ¼ 0.167). * Significant difference of postoperative bone defect area between the RP and conventional group is identified (p ¼ 0.02). Defect area change (%) was 18.7% in the conventional group and 8.03% in the RP group, and these values showed significant difference (p < 0.01).
patients in the conventional group had complete resolution at 6 month follow up, and the other 2 patients in the conventional group had residual diplopia that required revisional operation (Table 4). 4. Discussion The challenges associated with the management of orbital wall fractures originate from the complex skeletal and soft tissue anatomy of the orbit. It is important to consider the unique contours of the orbital anatomy, and this anatomical region must be precisely corrected to reconstitute the orbital volume and contour (Gordon et al., 2012). Incorrect reconstruction of the orbital dimensions, whether under- or over-corrected, may cause severe posttraumatic orbital deformities such as enophthalmos, diplopia, and
visual acuity disturbance (Manolidis et al., 2002; Shin et al., 2013). For reconstruction of medial wall and orbital floor fractures, the implants should be prepared to fit this exact contour, but this goal is difficult to achieve by free-hand implant bending. Given that the implant may need to be inserted and removed two or three times before the best shape and orientation are determined, this can place the periorbital soft tissues at increased risk of iatrogenic injury. Without additional tools such as intraoperative imaging or navigation, the placement of the implants is likely to be incorrect (Schon et al., 2006; Stoetzer et al., 2011). The posterior orbital floor and medial orbital wall are the key areas of the orbit with their own unique contours. The contour of the orbital floor slopes down directly behind the infraorbital rim, before ascending toward the posterior aspect, resulting in an Sshaped orbital floor in the posterior third of the orbit. When
Table 3 Comparisons of postoperative layout angle and gap length for medial and inferior wall fractures.
Medial wall fracture RP group conventional group Inferior wall fracture RP group conventional group a b
n
Layout angle (degree, Mean ± SD)
p value
Gap length (mm, Mean ± SD)
p value
24 22
3.49 ± 1.97 9.03 ± 4.9
p < 0.01a
3.1 ± 1.46 6.37 ± 3.29
p < 0.01b
20 16
2.23 ± 1.37 4.69 ± 2.51
p < 0.01a
5.21 ± 2.67 6.64 ± 3.61
p ¼ 0.12
Significant difference of layout angle was identified between the two groups. Significant difference of gap length was identified between the two groups.
934
Y.C. Kim et al. / Journal of Cranio-Maxillo-Facial Surgery 45 (2017) 928e936
Fig. 6. Comparison of Bone defect area between the RP and conventional groups for inferior wall fractures. y No statistical significance of the preoperative bone defect area between the RP and the conventional group (p ¼ 0.54). * Significant difference of postoperative bone defect area between the RP and conventional group is identified (p ¼ 0.01). Defect area change (%) was 12.8% in the conventional group and 7.14% in the RP group and these values showed significant difference (p < 0.01).
moving from the orbital floor to the medial orbital wall, the orbital floor inclines superiorly to meet the medial wall, creating a bony convergence in the posteromedial orbit. Restoration of this unique shape of the orbit following traumatic injury may often be difficult due to the inability of the surgeon to obtain adequate visibility and to verify proper implant placement during the operation (Ozer et al., 2016). One of the greatest advantages of using the RP skull model is that it permits direct visualization of the orbit, allowing surgeons to customize the patient-specific implants and simulate the placement of the implant onto the premorbid model. Most surgeons hesitate to dissect the posterior space of the orbit, in which optic nerve and ophthalmic vessels can be damaged. With the guide of RP skull model, surgeons are able to predict how far the implant can be inserted, where the exact location of the defect is, thereby achieving sufficient coverage of the defect even at the posterior side of the orbit (Fig. 7). For precise results, the orbital space can be volumetrically assessed using computer software. We described, in our previous reports of 127 traumatic orbital reconstructions, that precise volumetric reconstruction could be achieved using the mirrorimaged RP skull model (Park et al., 2015). Considering the clinical conditions, however, it is hard to access such software. 2-
Table 4 Clinical outcomes at postoperative 3 month visit in each group. Sign/symptom
RP group (n ¼ 44)
Enophthalmos None 95.45% (42/44) Mild 4.54% (2/44) Moderate 0 Severe 0 Presence of diplopia 4.54% (2/44) Limitation of ocular motility 0 97.72% (43/44) 1 2.27% (1/44) 2 0 3 0 4 0
Conventional group (n ¼ 38) 86.84% (33/38) 7.89% (3/38) 2.63% (1/38) 2.63% (1/38) 13.15% (5/38) 84.21% (32/38) 10.52% (4/38) 5.26% (2/38) 0 0
dimensional CT scans can also provide reliable information on the extent of the fracture and thereby can be used more practically for clinical situations (Ploder et al., 2002). We used three parameters to compare the surgical outcomes, and the overall achievement was much improved with the use of the RP skull models. The computed tomography parameters in our present study were easy to use and had some clinical value. The degree of enophthalmos could be predicted based on the area of the bone defect. Sung et al. (2013) suggested that an area of a bone defect of approximately 2.75 cm2 is likely to correspond to 2 mm of enophthalmos. In our current study, although the area of the bone defect was corrected properly in both groups, the mean reduction in the area of the bone defect was significantly better in the RP skull model group. This is most likely because the implant was molded as closely as possible to the original shape by direct visualization of the posterior bony shelf considering its unique bulging contour where it would have been limited to access it normally by conventional methods. It was possible to correlate the layout angle to the degree of enophthalmos in incorrectly placed implants as, even with a minimal change, it could cause a relatively large volumetric discrepancy in the original orbit. The orbital volumes after an implant insertion are likely to be smaller than the premorbid volumes. This can be explained by internal orbital placement of the implant and because the added thickness of the mesh result in an overall reduction in the orbital volume (Strong et al., 2013). Thus, the correct axis of the implant placement can be guaranteed by repetitive trials using the RP skull models before an implant insertion. Further analysis of the relationship between the layout angle and enophthalmos might suggest the ideal axis of implant position that would not increase enophthalmos. More recent developments in computer-aided design/ manufacturing have paved the way for patient-specific implants that provide more accurate restoration of the internal orbit and minimize the need for major intraoperative manipulations (Metzger et al., 2006). Although patient-specific implants seem to offer the best reconstructive contour, their fabrication is expensive and time-consuming (Strong et al., 2013). To date, 3D printed titanium orbital implants can cost between $ 3,000 and $ 5,000 (Vehmeijer et al., 2016). Otherwise, the costs of RP skull models
Y.C. Kim et al. / Journal of Cranio-Maxillo-Facial Surgery 45 (2017) 928e936
935
Fig. 7. Supportive role of using RP skull model when covering the defect at the posterior side of orbit. (a, b) Preoperative CT scans for medial wall and inferior wall fracture, respectively. (c, d) Postoperative CT scans showed the bone defect in the posterior orbit (arrows) can be sufficiently covered with the implants under guidance of RP skull model.
ranged from $ 300 and $ 500. Free-hand bent implants are more cost-effective but have lower contour accuracy and ease of use (Strong et al., 2013). There is no international consensus on the ideal material to be used for orbital reconstruction, however, the success of orbital fracture repair depends not only on the use of material that restores the premorbid orbital volume and shape, but also the surgeon's capacity to visualize the geometry of the bony defect spatially and to place the implants to fit the original contour. We felt that 3D-printed RP skull models could supply the geometric information of the fractured orbit and allowed the anatomical placement of implant materials until they fitted the defect as precisely as possible. The RP skull model is influenced by many factors affecting the stereolithographic process, therefore artifacts can occur because of poor equipment, patient movement, or the presence of metallic implants. Careful patient positioning and radiographic techniques can minimize these problems (Chang et al., 2003). Given the occurrence of artifacts that fail to reflect the thin orbital wall, especially through the lamina papyracea, CT slices less than 1.5 mm in thickness can minimize these errors. We hypothesized that the use of RP skull models would reduce the time of surgery by eliminating time spent on molding and repetitive fitting. However, there was no significant difference in operation time between the RP and conventional groups. 5. Conclusion More accurate restoration of traumatic orbital wall fractures can be achieved using patient-specific 3D-printed RP skull models.
Surgical outcomes assessed using computed tomography parameters show better results for RP skull model groups than models molded with a free-hand technique. Conflict of interest The authors have no financial interest in any of the products or devices mentioned in this article. Acknowledgments All persons who have contributed to the study are listed as authors, since everyone has met the listed criteria for authorship. There exist no current funding sources for this study. References Chang PS, Parker TH, Patrick Jr CW, Miller MJ: The accuracy of stereolithography in planning craniofacial bone replacement. J Craniofac Surg 14: 164e170, 2003 Gordon CR, Susarla SM, Yaremchuk MJ: Quantitative assessment of medial orbit fracture repair using computer-designed anatomical plates. Plast Reconstr Surg 130: 698ee705e, 2012 Lee KM, Park JU, Kwon ST, Kim SW, Jeong EC: Three-dimensional pre-bent titanium implant for concomitant orbital floor and medial wall fractures in an East asian population. Arch Plast Surg 41: 480e485, 2014 Manolidis S, Weeks BH, Kirby M, Scarlett M, Hollier L: Classification and surgical management of orbital fractures: experience with 111 orbital reconstructions. J Craniofac Surg 13: 726e738, 2002 Metzger MC, Schon R, Weyer N, Rafii A, Gellrich NC, Schmelzeisen R, et al: Anatomical 3-dimensional pre-bent titanium implant for orbital floor fractures. Ophthalmology 113: 1863e1868, 2006 Ozer MA, Govsa F, Kazak Z, Erdogmus S, Celik S: Redesign and treatment planning orbital floor reconstruction using computer analysis anatomical landmarks. Eur Arch Otorhinolaryngol 273: 2185e2191, 2016
936
Y.C. Kim et al. / Journal of Cranio-Maxillo-Facial Surgery 45 (2017) 928e936
Park SW, Choi JW, Koh KS, Oh TS: Mirror-imaged rapid prototype skull model and pre-molded synthetic scaffold to achieve optimal orbital cavity reconstruction. J Oral Maxillofac Surg 73: 1540e1553, 2015 Ploder O, Klug C, Backfrieder W, Voracek M, Czerny C, Tschabitscher M: 2D- and 3Dbased measurements of orbital floor fractures from CT scans. J Craniomaxillofac Surg 30: 153e159, 2002 Schon R, Metzger MC, Zizelmann C, Weyer N, Schmelzeisen R: Individually preformed titanium mesh implants for a true-to-original repair of orbital fractures. Int J Oral Maxillofac Surg 35: 990e995, 2006 Shin JW, Lim JS, Yoo G, Byeon JH: An analysis of pure blowout fractures and associated ocular symptoms. J Craniofac Surg 24: 703e707, 2013 Stoetzer M, Rana M, von See C, Eckardt AM, Gellrich NC: Reconstruction of defects of maxillary sinus wall after removal of a huge odontogenic lesion using prebended 3D titanium-mesh and CAD/CAM technique. Head Face Med 7: 21, 2011
Stoor P, Suomalainen A, Lindqvist C, Mesimaki K, Danielsson D, Westermark A, et al: Rapid prototyped patient specific implants for reconstruction of orbital wall defects. J Craniomaxillofac Surg 42: 1644e1649, 2014 Strong EB, Fuller SC, Wiley DF, Zumbansen J, Wilson MD, Metzger MC: Preformed vs intraoperative bending of titanium mesh for orbital reconstruction. Otolaryngol Head Neck Surg 149: 60e66, 2013 Sung YS, Chung CM, Hong IP: The correlation between the degree of enophthalmos and the extent of fracture in medial orbital wall fracture left untreated for over six months: a retrospective analysis of 81 cases at a single institution. Arch Plast Surg 40: 335e340, 2013 Vehmeijer M, van Eijnatten M, Liberton N, Wolff J: A novel method of orbital floor reconstruction using virtual planning, 3-dimensional printing, and autologous bone. J Oral Maxillofac Surg 74: 1608e1612, 2016