Computer-Assisted Design and Rapid Prototype Modeling in Microvascular Mandible Reconstruction

Transcription

1 The Laryngoscope VC 2012 The American Laryngological, Rhinological and Otological Society, Inc. Computer-Assisted Design and Rapid Prototype Modeling in Microvascular Mandible Reconstruction Matthew M. Hanasono, MD; Roman J. Skoracki, MD Objectives/Hypothesis: To evaluate the use of computer-assisted design and rapid prototype modeling to improve the speed and accuracy of mandibular reconstruction. Study Design: Case-control study. Methods: Between 2005 and 2011, 38 subjects underwent fibula free flap mandibular reconstruction using computerassisted design and rapid prototype modeling. Titanium plates were prebent using the models prior to surgery. Direct plate bending on the native mandible to accurately restore occlusion would not have been possible in 11 patients with exophytic tumors, nine patients with pathologic fractures, and 10 patients with a prior segmental mandibulectomy. Computer-generated cutting guides were utilized to facilitate fibular osteotomies. Results: The mean operative time for subjects was hours compared to the mean operative time defectmatched control group, for whom computer-assisted design and models were not used, of hours (P ¼.0006). Comparison of the preoperative and postoperative mandibles demonstrated that the mean change in position of selected bony landmarks (condyles, gonions, and gnathion) was less in the subject group than in the control group ( mm vs mm, respectively; P ¼.001) Comparison of postoperative mandibles with preoperative virtual plans showed a mean deviation of mm from planned fibular segment lengths and from planned angles between fibular segments. Conclusions: Computer-assisted design and rapid prototype modeling have the potential to increase the speed and accuracy of mandibular reconstruction. We believe these technologies are particularly useful for cases in which the original architecture of the mandible has been distorted or destroyed. Key Words: Mandibular reconstruction, fibula free flap, rapid prototype modeling, computer-assisted design. Level of Evidence: 3b Laryngoscope, 123: , 2013 From the Department of Plastic Surgery, The University of Texas M. D. Anderson Cancer Center, Houston, Texas, U.S.A. Editor s Note: This Manuscript was accepted for publication August 10, Presented at the Plastic Surgery Research Council 2010 Annual Meeting, San Francisco, California, U.S.A., May The authors have no funding, financial relationships, or conflicts of interest to disclose. Send correspondence to Matthew M. Hanasono, MD, Department of Plastic Surgery, University of Texas M. D. Anderson Cancer Center, 1515 Holcombe Boulevard, Unit 443, Houston, TX mhanasono@mdanderson.org DOI: /lary INTRODUCTION Advances in computer-assisted design (CAD) software have resulted in the ability to manipulate three-dimensional representations of the facial skeleton to perform virtual reconstruction of missing or abnormal parts. 1,2 Life-size physical models of the reconstructed facial skeleton can then be created by specialized printers using a technique known as rapid prototype modeling (RPM). 3 5 Recently, there has been growing enthusiasm for using these technologies to plan mandibular reconstruction with vascularized bone flaps. 6 8 Besides assisting with surgical planning, rapid prototype models can be used as templates to bend titanium hardware needed for vascularized bone flap fixation prior to surgery Furthermore, custom cutting guides can be manufactured using RPM technology to help make osteotomies at the appropriate lengths and angles to match the virtual plan. 13,14 We describe our protocol for shaping fibula free flaps virtually using CAD software, prebending titanium plates based on rapid prototype models, and performing osteotomies using computer-generated cutting guides. Our hypothesis was that CAD and RPM saves operative time and increases accuracy in microsurgical mandibular reconstruction, which remains one of the most arduous and challenging surgeries performed in head and neck cancer patients. To test this hypothesis, we compare outcomes in patients reconstructed with the aid of CAD and RPM to outcomes in defect-matched patients who underwent mandibular reconstruction without these tools. This report describes the largest experience in using CAD and RPM to assist with mandibular reconstruction to date and is the first to compare outcomes in reconstructions using these technologies to reconstructions using conventional techniques. MATERIALS AND METHODS Technique A digital three-dimensional model of the maxillofacial skeleton was created using fine-cut (1.5 mm) axial computed tomography (CT) data saved in Digital Imaging and Communications in 597

2 Fig. 2. Computer-generated cutting guide used to make fibular osteotomies at the lengths and angles required to replicate the virtually planned reconstruction. Fig. 1. Rapid prototype model based on a computer-assisted design reconstruction of a mandible using a fibula free flap. A titanium reconstruction plate was bent along the contours of the model preoperatively to save operative time. Medicine format. A virtual mandibular reconstruction with a fibula bone was performed using SurgiCase CMF (Materialise NV, Leuven, Belgium) software. The mandibular resection was estimated, taking into consideration the extent of the lesion as well as feedback from the resecting surgeon. The fibular osteotomies were planned by visualizing the reconstruction superimposed on the preoperative image of the mandible such that the outer (inferior-lateral mandibular border) contour of the mandible was restored. For cases in which a portion of the mandible was already absent, comparison to the normal side and the opposing maxillary dentition facilitated surgical planning. A standardized virtual fibula was used in this series, but the patient s own fibula may be utilized if a lower extremity CT scan is performed. Physical models were then created using RPM technology. As mentioned, RPM involves creation of a solid object that results in a highly accurate three-dimensional model of the mandible. RPM printing technology is based on layer-by-layer deposition of liquid polymer that is cured by laser or ultraviolet light, or starch powder bonded with glue, depending on the type of printer, into a solid model that is reported accurate from 1.5 mm to as high as mm. 3,5,10 Titanium reconstruction plates were then bent preoperatively to match the contours of the fibula and adjoining facial bones on the rapid prototype model (Fig. 1). The rapid prototype models were then sterilized for intraoperative use as a reference. A computer-generated cutting guide was also manufactured using the same RPM process to assist in making the osteotomies needed to shape the fibula bone. During surgery, the sterilized cutting guide was temporarily fixed to the harvested fibula bone using 10-mm monocortical screws, and a reciprocating saw blade was inserted into slots in the cutting guide to make osteotomies at the lengths and angles required to replicate the virtual plan (Figs. 2 and 3). A third party (Medical Modeling Inc., Golden, CO) performed the CAD manipulations and created the rapid prototype models for all subjects in this series. free flap was used for reconstruction, alone or in combination with another flap, when the defect included extensive involvement of oral or facial soft tissues. To help draw conclusions regarding surgical time savings, potential complications associated with this technique, and improvements in reconstructive accuracy, a defect-matched control group was selected from 183 other patients who underwent mandibular reconstruction using a fibula free flap performed by the authors during the same time period without CAD and RPM. In control cases, titanium reconstruction plates were bent along the contours of the native mandible, whenever possible, then used to guide fibular osteotomies. In cases where the native mandible was destroyed, malpositioned, or substantially distorted by tumor, the resection was performed, and the plate was bent in such a way as to best maintain centric occlusion of the remaining dentition as estimated by the surgeon. Patients were also matched with respect to whether they received a second flap as well as the type of flap (soft tissue free flap or pectoralis major myocutaneous pedicled flap). The accuracy of the reconstructions were compared by superimposing the preoperative image of the preresection mandible onto the postoperative image of the reconstructed mandible, both obtained from reformatted CT scan data whenever both were available. Patients were excluded from this Data Analysis Subjects included 38 patients who underwent microvascular free flap reconstruction after segmental mandibular resection utilizing CAD and RPM performed by the authors (M.M.H. and R.J.S.) between 2005 and In all cases, a fibula 598 Fig. 3. A side-by-side comparison of a rapid prototype model and the completed fibula free flap reconstruction.

3 analysis if the preoperative mandible was absent, for example, due to a prior resection, or misaligned in such a way that a comparison with the reconstructed mandible was not feasible, for example, if the patient had a displaced pathologic fracture. This process involved lining up the unaltered portions of the mandible such that the maximum overlap was achieved. The absolute difference in distances between the preoperative location and postoperative location of five bony landmarks (left and right condyle, left and right gonion, and the gnathion) was measured in subjects and in controls and compared with the assumption that smaller differences were consistent with a more accurate mandibular reconstruction (Fig. 4). As another way to measure the accuracy achieved using CAD and RPM, the actual postoperative length of each fibula segment, again based on postoperative CT data, was compared to the planned length of the same fibular segment and measured along the inferior-lateral border in subjects whenever postoperative CT scans were available. Similarly, the actual angles between each two fibular segments were compared to the planned angles between the same two fibular segments in the same subjects (Fig. 5). The mean differences between the actual and the planned segment lengths and angles were then calculated. If a given segment or angle was purposefully altered or omitted by the surgeon from the virtually planned reconstruction because the extent of the resection changed intraoperatively, such as in the case of tumor growth or regression, or a positive pathologic margin, then that segment length or angle was omitted from the analysis. Continuous data (reported as mean 6 standard deviation) were compared using the unpaired t test if the data were normally distributed or the Wilcoxon rank sum test if the data were nonparametric. Categorical data were compared using the v 2 test or Fisher exact test, as appropriate. All tests were two-tailed. P values <.05 were considered significant. Institutional review board approval was obtained prior to undertaking this study. Fig. 5. A comparison of the virtually planned (A) and actual postoperative (B) fibular lengths and osteotomy angles used as a second method to assess the accuracy of reconstruction. Fig. 4. A comparison of the positions of preoperative (black dots) and postoperative (gray dots) mandibular bony landmarks used to assess the accuracy of reconstruction. RESULTS The study population included 26 males and 12 females, with a mean age of years. Diagnoses included: osteoradionecrosis (n ¼ 14), squamous cell carcinoma (n ¼ 10), osteosarcoma (n ¼ 4), ameloblastoma (n ¼ 5), bisphosphonate necrosis (n ¼ 2), adenoid cystic carcinoma (n ¼ 1), melanoma (n ¼ 1), and gunshot trauma (n ¼ 1). In addition to fibula free flaps, 19 patients had a second flap to provide additional tissue for oral or facial soft tissue reconstruction, including 14 anterolateral thigh free flaps, one rectus abdominis myocutaneous free flap, and four pectoralis major myocutaneous pedicled flaps. The range of defects, based on anatomic divisions of the mandible, and number of fibula osteotomies performed for each reconstruction are summarized in Table I. Patient characteristics including prior treatments and medical comorbidites for the subject group that underwent mandibular reconstruction utilizing CAD and RPM and the control group that did not are shown in Table II. There were no significant differences in the prevalence of any of these characteristics between the two groups. The control group included 28 males and 10 females, with a mean age of years (P ¼.40). The mean time from the date of surgery to the 599

4 TABLE I. Defect Details in Patients (38 Subjects) Undergoing Mandibular Reconstruction Using Computer-Assisted Design and Rapid Prototype Modeling. TABLE III. Morphologic Findings in Patients Undergoing Mandibular Reconstruction With (38 Subjects) and Without (38 Controls) Computer-Assisted Design and Rapid Prototype Modeling. Defect Osteotomies Patients (%) Indication Subjects (%) Controls (%) P Value Condyle to contralateral angle 5 2 (5.3) Condyle to contralateral midbody 4 2 (5.3) Condyle to symphysis 3 1 (2.6) Ramus to contralateral midbody 3 2 (5.3) Condyle to contralateral parasymphysis 3 6 (15.8) Angle to contralateral angle 3 3 (7.9) Angle to contralateral midbody 3 3 (7.9) Angle to contralateral parasymphysis 2 5 (13.2) Mid-body to contralateral midbody 2 1 (2.6) Mid-body to contralateral parasymphysis 2 1 (2.6) Condyle to ipsilateral parasymphysis 2 2 (5.3) Ramus to ipsilateral parasymphysis 2 1 (2.6) Angle to symphysis 1 4 (10.5) Angle to ipsilateral parasymphysis 1 3 (7.9) Ramus to ipsilateral midbody 1 1 (2.6) close of the study for the subjects was months, and the mean time from the date of surgery to the close of the study for the controls was months (P ¼.42). Additional findings related to the preoperative mandibular morphology that would otherwise make mandibular reconstruction more challenging are summarized in Table III, including large exophytic tumors precluding plate bending along the native mandibular contours, pathologic fractures with displaced mandibular segments, and prior resections with or without free flap reconstruction. Such findings were significantly more common in the subject group (P <.0001). Complications are compared between the cases in which CAD and RPM were used and defect-matched controls in Table IV. There were no significant differences in recipient site complications, donor site complications, or TABLE II. Characteristics of Patients Undergoing Mandibular Reconstruction With (38 Subjects) and Without (38 Controls) Computer-Assisted Design and Rapid Prototype Modeling. Characteristic No. Subjects (%) No. Controls (%) P Value Prior surgery 20 (52.6) 12 (31.6).10 Prior radiation 21 (55.3) 18 (47.4).65 Prior chemotherapy 12 (31.6) 7 (18.4).29 Tobacco use 12 (31.6) 11 (28.9) 1.0 Alcohol use 10 (26.3) 14 (36.8).46 Medical comorbidities Cardiac disease 6 (16.2) 4 (10.8).73 Hypertension 12 (32.4) 17 (45.9).34 Cerebrovascular disease 2 (5.4) 3 (8.1) 1.0 Pulmonary disease 3 (8.1) 3 (8.1) 1.0 Diabetes mellitus 4 (10.8) 4 (10.8) 1.0 Large exophytic tumor 11 (28.9) 3 (7.9).03 Pathologic fracture 9 (23.7) 3 (7.9).11 Prior resection 5 (13.2) 2 (5.3).43 (free flap reconstruction) Prior resection 5 (13.2) 1 (2.7).20 (plate reconstruction) Total 30 (78.9) 9 (23.7) <.0001 medical complications. The mean follow-up time was months for the subject group and months for the control group. Functionally, all patients in both groups were decannulated of their tracheostomies, and all patients in both groups whose defect was isolated to the mandible (i.e., single free flap patients) were able to tolerate a mechanical soft or regular diet. Operative times are compared between the group that underwent mandibular reconstruction using CAD and RPM and the control group (Table V). Operative times in cases requiring a fibula free flap for reconstruction alone and cases requiring a fibula free flap and a second free or pedicled flap were analyzed separately. Mean operative times were shorter in both types of cases, although the operative time was only significantly shorter in single free flap cases. In 12 subjects and 16 controls, postoperative threedimensional mandibular images were superimposed on TABLE IV. Complications in Patients Undergoing Mandibular Reconstruction With (38 Subjects) and Without (38 Controls) Computer-Assisted Design and Rapid Prototype Modeling. Complication Subjects (%) Controls (%) P Value Recipient site 8 (20.1) 9 (23.7).78 Free flap venous congestion 3 (7.9) 1 (2.7) Wound dehiscence 2 (5.3) 3 (8.1) Infection 2 (5.3) 5 (13.5) Hematoma 2 (5.3) 0 (0) Free flap loss 1 (2.6) 1 (2.6) Fistula 0 (0) 1 (2.6) Donor site 6 (15.8) 8 (20.1).77 Wound dehiscence 3 (8.1) 3 (7.9) Infection 1 (2.6) 3 (7.9) Partial skin graft loss 1 (2.6) 2 (5.3) Hematoma 1 (2.6) 0 (0) Medical 3 (7.9) 4 (10.5) 1.00 Pneumonia 2 (5.3) 2 (5.3) Atrial fibrillation 1 (2.6) 0 (0) Myocardial infarctions 0 (0) 1 (2.6) Acute cholecystitis 0 (0) 1 (2.6) Total* 15 (39.5) 17 (44.7).82 *Five subjects and five controls had more than one complication each. 600

5 TABLE V. Operative Times in Patients Undergoing Mandibular Reconstruction With a Fibula Free Flap With (38 Subjects) and Without (38 Controls) Computer-Assisted Design and Rapid Prototype Modeling. Operative Time Subjects (Hours 6 SD) Controls (Hours 6 SD) P Value Single flap cases Double flap cases SD ¼ standard deviation. preoperative three-dimensional mandibular images. Table VI summarizes the differences between preoperative and postoperative mandibular morphology based on the location of five selected bony landmarks. Additionally, the lengths of each fibular segment and angles between fibular segments following surgery were compared to the virtually planned lengths and angles in 18 subjects who underwent reconstruction using the CAD and RPM protocol (Table VII). Eight of 55 possible segments and five of 38 possible angles were omitted from the analysis because the surgeon purposefully changed the length or number of segments due to intraoperative findings. Overall, 42.1% of subject cases involved some deviation from the preoperative plan, although only 17.8% of all fibular segments required lengthening, shortening, or omission due to the extent of resection changing based on intraoperative findings, such as interval tumor growth or shrinkage due to neoadjuvant chemotherapy. A representative case involving a 61-year-old man with bilateral mandibular osteoradionecrosis is presented in Figures 6 to 10. Use of CAD allowed us to plan the reconstruction such that the length of fibula needed was limited to 26.9 cm, which coincided with the maximum length of bone available from the patient s lower extremity. TABLE VII. Mean Difference Between Planned and Actual Fibular Segment Lengths and Osteotomy Angles in Patients (18 Subjects) Undergoing Mandibular Reconstruction With Computer-Assisted Design and Rapid Prototype Modeling. Measurement Sample Size (n) Difference (6 SD) Fibular segment length mm Fibular osteotomy angle SD ¼ standard deviation. most accurately shape vascularized bone flaps so that facial symmetry as well as function are best restored. Minimizing operative time by maximizing the efficiency of these complex surgeries is another important goal. In the present study, we utilized CAD and RPM to assist with microvascular mandibular reconstruction. Patient characteristics that might influence outcomes were very similar to those in a defect-matched control group who underwent fibula free flap mandibular reconstruction by the conventional method. However, subjects who underwent reconstruction with the aid of CAD and RPM were more likely to have mandibles with large exophytic tumors, pathologic fractures, or surgically absent mandibular segments than the control group, potentially making their reconstructions more challenging if anything because shaping the reconstruction along the contours of the native mandible would be difficult or impossible. In terms of outcomes, we did not find a significant difference in the complication rates observed between the two groups, suggesting that use of this protocol is not associated with any potential hazards to patient safety. However, operative time was reduced both for fibula free flap-only cases and for multiple flap cases (although statistically significant only in the single free flap cases). We ascribe the time savings to prebending DISCUSSION Mandibular reconstruction has changed significantly due to the introduction of vascularized bone flaps, and now very extensive defects can be reconstructed. One of the greatest challenges that remains is how to TABLE VI. Mean Difference in Position of Bony Landmarks Between Preoperative and Postoperative Mandibles in Patients Undergoing Mandibular Reconstruction With a Fibula Free Flap With (12 Subjects) and Without (16 Controls) Computer-Assisted Design and Rapid Prototype Modeling. Landmark Subjects (mm 6 SD) Controls (mm 6 SD) P Value Left condyle Left gonion Gnathion Right gonion Right condyle Cumulative (all points) SD ¼ standard deviation. Fig. 6. Computed tomography image from a 61-year-old patient with bilateral pathologic fractures of the mandible due to osteoradionecrosis. 601

6 Fig. 7. Patient with bilateral pathologic fractures of the mandible due to osteoradionecrosis with an open bite (A) and an orocutaneous fistula near the angle of the left mandible (B). hardware and use of the cutting guides to simplify osteotomies. The preoperative planning that occurs may have also increased operative efficiency by improving surgical decision making and reducing trial and error during fibular harvest and shaping. Fig. 8. Virtual reconstruction of a planned near-total mandibular defect following debridement of the necrotic mandible. 602 Although the cost of CAD and RPM services is not negligible (at time of article preparation, charges for CAD and RPM services range from $3,000 to $4,700), a reduction in operating room time may help mitigate some of financial burden of using these technologies. 15 More important potential consequences of time-savings associated with use of CAD and RPM would include a decrease in surgeon fatigue, a source of burnout among microvascular surgeons, 16 as well as reduced surgical errors 17,18 and complications such as surgical site infections and venous thromboembolism, 19,20 which are associated with increased surgical time. We would argue that an average of 1.7 hours of surgical time saved in surgeries that typically last 8 to 10 hours or more is meaningful; however, we acknowledge that a threshold for precisely how much time savings in microvascular reconstructive surgery is required to achieve such benefits has not been established. In addition to examining operative time savings, we attempted to examine the accuracy of reconstruction using two methods. In the first method, we compared the postoperative location of five bony landmarks to their preoperative location in three-dimensional space. Minimizing deviation of these bony landmarks from their native positions was considered a measure of reconstructive accuracy. We found that, for all five landmarks, mandibles reconstructed using CAD and RPM had a mean deviation that was less than mandibles reconstructed using the conventional method. Although the difference between subjects and controls was not significant for any given landmark, likely due to the limited sample size, the differences between all measurements combined was statistically significant.

7 Fig. 9. The reconstruction in Figure 8 was planned such that all of the available fibular bone in this patient could be used. As another means of measuring accuracy, we compared the actual fibular segment lengths and the angles between segments created with the aid of computer-generated cutting guides to the lengths and angles judged to be optimal during preoperative virtual planning. We found that actual lengths deviated from the planned lengths by an average of 2.40 mm and that actual angles deviated from the planned angles by an average of In a similar study, Roser et al. 21 achieved an even smaller mean deviation of 1.30 mm between the planned and actual fibular segments. We ascribe some of this inaccuracy to the leeway allowed by the cutting guide slots as well as to imperfect bending or placement of the titanium plates by the surgeon. Although ideally we would have performed analyses of reconstructive accuracy on all subjects and controls in our series to increase the statistical power of our findings, postoperative CT data were sometimes lacking due to recurrent disease or lack of follow-up secondary to patients receiving continued care elsewhere. In some cases, preoperative CT data had to be excluded from the comparison because the mandible was grossly abnormal and bony landmarks were malpositioned. Additionally, incremental improvements in geometric accuracy, although intuitively worthwhile, still need to be proven to result in clinically useful increases in function, aesthetics, or patient satisfaction. Based on our preliminary data, we suspect that such questions will be answered favorably as multicenter experience with large numbers of subjects results from increased use of CAD and RPM in reconstructive surgery. An important limitation of CAD and RPM that we encountered was the potential for the extent of resection to change. Changes in the extent of resection may reduce the usefulness of this technique by requiring adjustments in bone segment length or number of osteotomies as well as altering the shape of the titanium plate, potentially Fig. 10. Postoperative appearance of the patient described in Figures 6 through 9: anterior-posterior view (A) and left lateral view (B). An anterolateral thigh free flap was also used for neck soft tissue reconstruction in addition to the fibula free flap mandible reconstruction. 603

8 increasing operative time and decreasing the accuracy of the reconstruction as it deviates from the preoperative plan. In our experience, it is often useful to tend toward modest overestimation of the defect because the plate cutting guides can still be useful even if a small amount of the fibula needs to be cut back or the mandibular resection is slightly enlarged. Use of recent imaging for virtual planning and close communication with the resecting surgeon can minimize changes from initial plan. CONCLUSION We present a method based on CAD and RPM that increases the accuracy of microvascular mandibular reconstruction while saving time by streamlining the operative procedure, simplifying the creation of osteotomies, and allowing for prebending of the titanium plates needed for flap fixation. We predict that the indications for CAD and RPM use in reconstructive surgery will continue to expand as the technology evolves and favorable clinical outcomes data are generated from further experience. Based on our initial experience, we recommend this reconstructive technique for all complicated mandibular reconstructions, particularly in patients with distorted, fractured, or missing mandibular segments. Acknowledgments The authors thank Ms. Katie A. Weimer, Mr. Kurt Moore, and Mr. Andrew M. Christensen of Medical Modeling, Incorporated for their invaluable assistance, including creating all computer images and medical models for this article. BIBLIOGRAPHY 1. Eckardt A, Swennen GR. Virtual planning of composite mandibular reconstruction with free fibula bone graft. J Craniofac Surg 2005;16: Bell RB. Computer planning and intraoperative navigation in cranio-maxillofacial surgery. Oral Maxillofac Surg Clin North Am 2010;22: Sinn DP, Cillo JE, Miles BA. Stereolithography for craniofacial surgery. J Craniofac Surg 2006;17: Choi JY, Choi JH, Kim NK, et al. Analysis of errors in medical rapid prototype models. Int J Oral Maxillofac Surg 2002;31: Chang PS, Parker TH, Patrick CW Jr, Miller MJ. The accuracy of stereolithography in planning craniofacial bone replacement. J Craniofac Surg 2003;14: Bill JS, Reuther JF, Dittman W, et al. Stereolithography in oral and maxillofacial operation planning. Int J Oral Maxillofac Surg 1995;24(1 pt 2): Ueda KM, Tajima SM, Oba SM, et al. Mandibular contour reconstruction with three-dimensional computer-assisted models. Ann Plast Surg 2001; 46: Kernan BT, Wimsatt JA. Use of a stereolithography model for accurate, preoperative adaptation of a reconstruction plate. J Oral Maxillofac Surg 2000;58: Juergens P, Krol Z, Zeilhofer HF, et al. Computer simulation and rapid prototyping for the reconstruction of the mandible. J Oral Maxillofac Surg 2009;67: Cohen A, Laviv A, Berman P, Nashef R, Abu-Tair J. Mandibular reconstruction using stereolithographic 3-dimensional printing modeling technology. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 2009;108: Bell RB, Weimer KA, Dierks EJ, Buehler M, Lubek JE. Computer planning and intraoperative navigation for palatomaxillary and mandibular reconstruction with fibular free flaps. J Oral Maxillofac Surg 2011;69: Antony AK, Chen WF, Kolokythas A, Weimer KA, Cohen MN. Use of virtual surgery and stereolithography-guided osteotomy for mandibular reconstruction with the free fibula. Plast Reconstr Surg 2011;128: Hirsch DL, Garfein ES, Christensen AM, Weimer KA, Saddeh PB, Levine JP. Use of computer-aided design and computer-aided manufacturing to produce orthognathically ideal outcomes: a paradigm shift in head and neck reconstruction. J Oral Maxillofac Surg 2009;67: Hanasono MM, Jacob RF, Bidaut L, Robb GL, Skoracki RJ. Mid-facial reconstruction using virtual planning, rapid prototype modeling, and stereotactic navigation. Plast Reconstr Surg 2010;126: Macario A. What does one minute of operating room time cost? J Clin Anesthesia 2010;22: Contag SP, Golub JS, Teknos TN, et al. Professional burnout among microvascular and reconstructive free-frlap head and neck surgeons in the United States. Arch Otolaryngol Head Neck Surg 2010;136: Sturm L, Dawson D, Baughan R, et al. Effects of fatigue on surgeon performance and surgical outcomes: a systematic review. ANZ J Surg 2011: 81: Veasey S, Rosen R, Barzansky B, Rosen I, Owens J. Sleep loss and fatigue in residency training. JAMA 2002;288: Greenblatt DY, Rajamanickam V, Mell MW. Predictors of surgical site infection after open lower extremity revascularization. J Vasc Surg 2011;54: Haridas M, Malangoni MA. Predictive factors for surgical site infection in general surgery. Surgery 2008;144: Roser SM, Ramachandra S, Blair H, et al. The accuracy of virtual surgical planning in free fibula mandibular reconstruction: comparison of planned and final results. J Oral Maxillofac Surg 2010;68: