Primary Dural Repair via an Endoscopic Endonasal Corridor: Preliminary Development of a 3D-Printed Model for Training

Abstract

Objectives  Endonasal suturing is an investigational method for dural repair that has been reported to decrease the incidence of cerebrospinal fluid fistula. This method requires handling of single-shaft instrumentation in the narrow endonasal corridor. In this study, we designed a low-cost, surgical model using three-dimensional (3D) printing technology to simulate dural repair through the endonasal corridor and subsequently assess the utility of the model for surgical training.

Methods  Using an Ultimaker 2+ printer, a 3D-printed replica of the cranial base and nasal cavity was fitted with tissue allograft to recapitulate the dural layer. Residents, fellows, and attending surgeons were asked to place two sutures using a 0-degree endoscope and single-shaft needle driver. Task completion time was recorded. Participants were asked to fill out a Likert scale questionnaire after the experiment.

Results  Twenty-six participants were separated into groups based on their prior endoscope experience: novice, intermediate, and expert. Twenty-one (95.5%) residents and fellows rated the model as “excellent” or “good” in enhancing their technical skills with endoscopic instrumentation. Three of four (75%) of attendings felt that the model was “excellent” or “good” in usefulness for training in dural suturing. Novice participants required an average of 11 minutes for task completion, as compared with 8.7 minutes for intermediates and 5.7 minutes for experts.

Conclusion  The proposed model appears to be highly effective in enhancing the endoscopic skills and recapitulating the task of dural repair. Such a low-cost model may be especially important in enhancing endoscopic facility in countries/regions with limited access to cadaveric specimens.

Introduction

The expanded endonasal approach (EEA) provides a minimally invasive corridor for direct access to the ventral cranial base while eliminating brain retraction. ¹ However, the high incidence of postoperative cerebrospinal fluid (CSF) fistula associated with its use remains a major limitation of the technique. Despite widespread introduction of the nasoseptal flap, the reported incidence of postoperative CSF fistula with EEAs has been cited to be as high as 20% in recent studies. ²

Several surgical techniques have been developed to address the high incidence of postoperative CSF fistula seen following EEAs. ³ Endonasal suturing is one such technique that has been investigated but has not yet gained widespread acceptance. ⁴ A small number of clinical reports support its use. Ahn and Kim reported the routine use of endonasal suturing in 21 patients, including 9 patients with craniopharyngiomas, who underwent transsphenoidal resection of suprasellar tumors. While the closure time in initial cases was impractical (reported as 9 hours in the first case), the incidence of postoperative CSF fistula in their study was zero. With additional practice and training, the time of closure dropped significantly to ∼1 hour in the final case reported. ⁵

While there are data to support the benefit of endonasal suturing in cases where tumor resection via EEA has resulted in widespread violation of the arachnoid layer, current training models and methods of closure are impractical. Thus, in this study, we sought to design an inexpensive, accessible, and reproducible model to replicate endonasal suturing of the dura of the cranial base in a training environment using advanced three-dimensional (3D) printing technology.

Materials and Methods

Model Design

A skull model was reproduced first by importing computed tomography (CT) scans in DICOM format into 3D Slicer (Slicer 4.10.2) for bone labeling. Thresholding was then achieved by manually highlighting the desired structures. The resultant virtual model of the viscerocranium and cranial base was then exported in STL format and uploaded into Meshmixer. The superior and middle conchae were removed to mimic a standard surgical approach into the sphenoid sinus. A 2 × 2.5 cm cranial base defect was created in the sella turcica and planum sphenoidale ( Fig. 1 ). The final STL file was transferred to Cura and primed for printing in polylactic acid using the Ultimaker 2+ 3D printer.

Fig. 1.

DICOM data of CT scans obtained for creation of 3D model of the skull using 3D Slicer, Meshmixer, and Cura programs. CT, computed tomography; 3D, three-dimensional.

Fresh chicken skin was used to replicate dura matter of the cranial base. To fixate the chicken skin to the model, five holes were made in the 3D skull around the cranial base defect using a 3.5-mm drill. Small corresponding openings were cut in the chicken skin, which was subsequently fixed using maxillofacial screws, providing complete coverage of the sellar and planum defects ( Fig. 2 ). Nail polish was used to designate optimal sites of needle passage for suturing ( Fig. 3 ), which in turn were used to plan the standard midline “dural” laceration.

Fig. 2.

Fresh chicken skin was fixed by screws to replicate dura mater.

Fig. 3.

Stepwise illustration of primary dural repair suturing via endoscopic endonasal corridor using 3D-printed model. ( A, B ) Participants were asked to insert the needle into the top right marked dot and exit the dural defect by passing underneath into the top left dot using one-hand technique. ( B ) With the needle tip exiting the contralateral side, the needle body is grasped and pulled out from the top left dot using a long-shaft forceps. ( C ) The needle body was inserted into the loop of the barbed suture inside the nasal cavity and was tightened by pulling upward outside the nasal cavity and released from the driver. ( D, E ) The needle body is placed back in the driver and the suturing steps were repeated in the same pattern on the bottom dots. ( F ) A completed endonasal dural suturing performed on the 3D-printed model. 3D, three-dimensional.

Investigation of Model

Both neurosurgery and otolaryngology residents and attending physicians volunteered to participate in the study. Prior to participation, an instructor demonstrated the required tasks. A midline laceration was created in the chicken skin by the instructor with an 11 blade. The participants were asked to repair this laceration with two throws of barbed suture (Stratofix 4.0 Knotless, Ethicon) using an endonasally compatible long-shaft microneedle driver (SEPEHRNIA Neurosurgical Micro Instruments; Karl Storz, Tuttlingen, Germany). Fresh chicken skin and new sutures were provided for each participant. Each participant used a Hopkins II 0 degree endoscope (Karl Storz) for visualization. As the barbed sutures were associated with an eyelet, all participants were required to insert the needle through the eyelet between the first and second throws. A detailed visual description of the technique utilized in this study is provided in Fig. 3 .

After successful completion of the task, the participants were asked to fill out one of two types of questionnaires in Likert scale format. Questionnaire A was provided to residents and fellows (Group A) and focused on perceived utility in various aspects of training. Questionnaire B was provided to attending physicians (Group B) who had a significant prior experience with endonasal procedures and focused on the ability of the simulated model to recapitulate the surgical experience. Both attending physicians and residents were queried on the usefulness of the model in training in endonasal suturing. Performance time was recorded for each participant. Two-sided t -tests (program R version 3.5.1.) were used to evaluate differences.

Results

Time

Residents, fellows, and attendings from otolaryngology and neurosurgery departments were invited to participate in the experiment. Total 26 participants successfully completed suturing on the model and were divided into novice (14), intermediate (8), and expert (4) groups for purposes of operative evaluation. The mean time suturing for all participants was 9.5 minutes ( Table 1 ). The mean time of experts was 5.7 minutes, which was 3 minutes faster than the 8.7 minutes mean time of intermediates ( p  = 0.3) and 5.3 minutes than the 11 minutes mean time of novices ( p  = 0.06).

Table 1. Task completion times.

	Average time to complete tasks (min)	Difference in time from expert group (min)
Expert	5.7	Reference
Intermediate	8.7	3 ( p  = 0.3)
Novice	11	5.3 ( p  = 0.06)
All	9.5

Note: Average time to complete tasks (minute) in novice, intermediate, and expert groups utilizing the 3D-printed model.

Questionnaire

After completing the tasks, participants were asked to evaluate the model through a Likert scale questionnaire to determine their subjective impression of the 3D model ( Table 2 ). The 26 total participants were divided into two groups for purposes of subjective assessment using Likert questionnaire. Group A consisted of 22 residents and fellows in training with <7 years of endoscopic experience. Group B consisted of two otolaryngologists and two neurosurgeons with 7 to 20 years of endoscopic experience.

Table 2. Posttask questionnaire responses.

Question/evaluation	Response
	Excellent (5)	Good (4)	Average (3)	Below average (2)	Poor (1)
Group A: residents and fellows ( n  = 22)
Usefulness in improving technical proficiency with endoscopy	15 (68.2%)	6 (27.3%)	1 (0.5)	0	0
Usefulness for training in dura suturing	10 (45.5%)	9 (41%)	3 (13.5)	0	0
Efficacy of the material in mimicking the dura	8 (36.4%)	7 (32%)	7 (32%)	0	0
Group B: attendings ( n  = 4)
Ability to replicate the technical steps of the surgical approach/closure	0	2	2	0	0
Usefulness for training in dura suturing	2	1	1	0	0

Note: Participant responses were gathered in Likert scale format with scores from 1 to 5.

Twenty-one (95.5%) individuals in Group A believed that the model was excellent or good in helping to improve technical proficiency with endoscopic instrumentation. Nineteen (86.5%) described the model as excellent or good in ability to improve dural suturing. However, only 15 (68.4%) felt the chicken skin was excellent or good at replicating dura mater. Only two out of the four (50%) attendings thought felt the simulated model was excellent or good in its ability to replicate the technical steps of the surgical approach and closure. Nevertheless, three (75%) believed that the model was excellent or good in its ability to improve technical proficiency with dural suturing.

Discussion

Dural suturing via a transsphenoidal corridor has previously been described as an effective technique for watertight closure of defects. ⁶ Kim et al ⁷ provided a retrospective analysis of 861 patients after pituitary adenoma resection and concluded that direct dural suturing should be routinely performed in the setting of wide dural defects or patients otherwise determined to be at high risk for development of postoperative CSF fistula. A recent cadaveric study by Kizmazoglu et al ⁸ provided scientific rational for dural suturing—demonstrating increased burst pressures in groups with dural suturing alone or suturing in combination with sealant or collagen graft, compared with groups without suturing. Horiguchi et al recommended fascia graft suturing, in combination with CSF, as an effective method to prevent CSF fistula after tumor resection resulting in high-flow CSF leak. ⁹ ·However, in the Horiguchi et al's study, creation of a watertight seal required securing the fascial graft with 15 to 20 knots, complicating the efficiency of the closure. Despite the aforementioned precedent supporting primary endoscopic repair of the dural layer after tumor resection associated with significant arachnoidal defect, ^{6 7 8 9} EDS is not routinely employed many areas of the world, including the United States. The authors believe that adequate surgical training on simulators could help improve surgical proficiency with EDS, making primary dural repair through an endonasal corridor a practical alternative.

EDS to facilitate primary repair of the dural layer through the endonasal corridor offers an intriguing solution to the significant problem of CSF fistula that has long plagued the EEA. It is clear that significant evolution in technology of single-shaft instrumentation and suturing material, as well as improved methods to master what is a technically demanding learning curve, will be required for widespread implementation of EDS. The proposed study sought to investigate an inexpensive and readily-available simulated model to help facilitate rapid acquisition of technical skills integral to EDS. For years, cadaveric training has remained the gold standard for surgical training—yet it is expensive, nonreplicable, and not universally available in the developing world. Virtual surgical models, including those that can be easily replicated with 3D printers, offer promising and cost-effective alternatives. ^{10 11} We feel the model proposed in this article provides a novel solution to address a well-defined problem.

While the use of 3D-printed material to replicate endoscopic training has been well described, ^{10 11 12 13 14 15} to the authors' knowledge, the use of simulated models to improve endoscopic dural suturing through an endonasal corridor has only been reported in a single previous article. Xie et al employed an acrylonitrile butadiene styrene plastic box with two 17-mm diameter holes, meant to recapitulate the nasal cavity and used an egg membrane for simulated dural suturing. ⁴ Our proposed model has several upgrades from the model proposed in the study by Xie et al: the anatomy and dimensions of the endonasal corridor, which are directly derived from CT imaging, are felt to be closer to the anatomy encountered in surgery. Additionally, the egg-shell membrane utilized in the Xie et al's study is likely to be uncharacteristically tenuous in comparison to normal human dura. We feel chicken skin is more realistic in this regard. As a final note, the model reported in the study by Xie et al used extracorporeal knot tying—an unnecessary encumbrance in the current surgical era. In contrast, our model employs the use of barbed suture, which obviates the need for cumbersome knot tying and tightening, and likely represents the future of endonasal dural repair. The clinical use of barbed suture for endoscopic dural repair has been previously described. ¹⁶

Preliminary investigation demonstrates the simulated model is not perfect in its current state. Only 15 (68.4%) of resident/fellow participants felt that chicken skin was “excellent” or “good” in simulating the dural layer. In this investigation, several materials were assessed as dural substitutes. Expanded polytetrafluoroethylene has been used in previous studies for dura mater replacement with promising results. ^{17 18} However, a preliminary investigation with our model demonstrated difficulty with needle insertion and suture retainment, which are central features needed in a training model. We also trialed both temporalis fascia and dura mater from embalmed specimens. Those materials were perhaps the most realistic with regard to desired thickness and handling characteristics. However, this tissue dried after a relatively short period of time, making it impractical for the proposed model. We ultimately settled on chicken skin, which is cheap, readily available, and maintains proper handling characteristics over long intervals. In addition to suboptimal feedback on dural simulation, only ∼50% of attendings felt the model was “excellent” or “good” in its ability to replicate the technical steps of the surgical approach/closure. It is possible in the future that additional time and planning could be dedicated to recreating more accurate soft tissue characteristics (e.g., inclusion of nasal tip; addition of mucosal, sinonasal, and sphenoidal neurovascular analogues) and anatomic restrictions characteristic of the endonasal corridor, which would be expected to improve this metric. However, the benefit of these additions should be weighed against costs of planning and technology necessary for implementation. The cost of the proposed model, excluding purchase of entry-level 3D printer, is less than 30 U.S. dollars. The model can be produced in less than 72 hours.

The majority (75%) of attendings felt the model was “excellent” or “good” in “usefulness for training in dural suturing.” Additionally, 95.5% of trainees believed that the model was “excellent” or “good” in helping to improve technical proficiency with endoscopic instrumentation; 86.5% of trainees described the model as “excellent” or “good” in ability to improve dural suturing. With these data in mind, the model appears to be better at training individuals how to perform primary dural repair through the endonasal corridor than recapitulating the technical steps of surgery. This is not entirely unexpected, as the primary intent of the 3D model was the former objective.

The surgical task assessed in this study was designed for a practical assessment of the utility of the proposed simulated model. Note is made that a more thorough assessment of the surgical proficiency of the participants in this study could be obtained with measurements, not only of time but also of surgical accuracy. While benefits of repetition were not specifically assessed in this study, we feel there would also be utility in assessing improvement as a function of repetition in future studies, as previous investigations of endoscopic microsurgery have helped with quantification of this correlation. ¹⁹ Intraoperative dural suturing in EEAs is fraught with several complicating factors that include, but are not limited to, the proximity of eloquent neurovascular structures to thin leaflets of partially resected dura and differences in planar angulation of the sella and planum sphenoidale. These complicating factors, which increase the technical difficulty associated with EDS, demand that any surgeon contemplating such a repair be facile with meticulous handling of single-shaft instrumentation and formatted suture material in the narrow and restricted endonasal corridor. The authors believe the simulated model proposed in this investigation represents a small but important step in overcoming these technical barriers.

Conclusion

The proposed simulated model, which can be easily produced with an entry-level 3D printer for low costs, was perceived by participants to be highly effective in ability to improve technical proficiency associated with endoscopic instrumentation and endonasal dural suturing. The model was less effective at recapitulating the technical steps of surgery and/or surgical anatomy. The authors feel the proposed simulated model represents a small but important step in improving technical facility with endonasal dural suturing.