Custom Laser-Cut Silastic^® Implants for Type I Thyroplasty in a Preclinical Model Free

Unilateral vocal fold paralysis (UVFP) is a condition characterized by impaired nerve function of the muscles of the vocal folds (VFs). This impairment frequently causes a gap between the VFs, which results in dysphonia and can negatively impact quality of life, employment, and productivity [1]. UVFP can cause dysphagia, which puts individuals at risk for aspiration pneumonia, malnutrition, or dehydration [1]. The standard of care for persistent UVFP is type I thyroplasty. Type I thyroplasty has been proven safe and effective for improving voice outcomes as well as reducing aspiration and improving deglutition [2, 3]. The procedure is reversible, and implants are customizable. Type I thyroplasty has demonstrated significant long-term improvements on vocal outcome measures such as stability of habitual intensity and maximum phonation time [4] and improves acoustic, aerodynamic, and perceptual voice outcomes [5].

Each patient’s laryngeal anatomy and tissue properties are unique and require a patient-specific approach to type I thyroplasty to attain optimal outcomes. Current best practice for individualized implants involves on-the-spot manual adjustments based on perceptual and qualitative assessments of the patient’s glottal space and voice quality [6]. Computational modeling to plan the surgical approach to UVFP will streamline the process of anatomically personalized implants with the aim to reduce the potential for early failures and complications.

The preclinical stage of research to develop a type I thyroplasty planning tool requires a method of producing implants of a precise size and shape made from medical-grade, bio-tolerable materials. Therefore, we used an iterative process to develop a standardized implant for this stage of the research process. Individualized implants may be further tailored to specifications informed by a computational model in the future.

3D modeling software (Tinkercad, 2019) was used to develop computer-aided design models of implants in two shapes and two sizes, for a combined total of four implant types. We designed two implant shapes: S1, rectangular, and S2, tetrahedral with greater depth oriented toward the posterior portion of the implant. Each shape was designed in two sizes: D1, 1 × 1 × 2 mm, and D2, 1 × 2 × 2 mm (Fig. 1a). Each implant had a shelf on the portion of the implant distal to VF placement to facilitate fit within the thyroplasty window. Implant size was scaled based on relative size differences between rabbit and human larynges. These parameters were input to a laser printer with a 50 W CO2 laser (Epilog Zing 24) at the University of Pittsburgh. Twelve implants of each size and shape (N = 48) were produced to determine production speed. The precision was set to 500 dpi. Implants were vector cut from a single sheet of medical-grade silicone elastomer (6 × 8 × 0.080 in non-reinforced sheeting, Bentec Medical, Woodland, CA, USA) at 50% power, 40% speed, and 5,000 Hz.

Five New Zealand white rabbits weighing between 2.5 kg and 4.0 kg were used in this study. Humane animal care and use in accordance with the Animal Welfare Act and the NIH Guide for the Care and Use of Laboratory Animals was reviewed, approved, and monitored by the University of Pittsburgh’s Institutional Animal Care and Use Committee (protocol #21120467). A non-stimulated in vivo rabbit phonation model as previously described [7‒9] was used to determine safety and physiological efficacy. Briefly, the cricoid and thyroid cartilages were drawn together with fixed sutures to narrow the glottal space in a phonatory configuration, and an incision was made in the trachea inferior to the larynx to pass humidified airflow through the glottis [10]. We performed type I thyroplasty using the Isshiki method [4, 11] and placed randomly selected implant shapes and sizes, specifically, S1D1 (n = 1), S1D2 (n = 1), S2D1 (n = 2), and S2D2 (n = 1). Stroboscopic laryngoscopy was used to monitor perioperative laryngeal geometry. Larynges were harvested, and 3D magnetic resonance imaging (MRI) was performed at 11.7 tesla with an isotropic resolution set at 60 μm.

Forty-eight implants were laser cut from Silastic^® sheeting in less than 1 min. Assessment of size and shape using calipers demonstrates reliability of the computer-aided design model (Fig. 1b). The approximate cost per implant is $0.30. Effective VF medialization was confirmed in vivo using laryngoscopic evaluation and ex vivo using MRI (Fig. 1c, d).

Model reconstructions of the larynx from MRI were performed to evaluate VF medialization. The larynx and VF cover were reconstructed before and after unilateral right-sided insertion of the implant (implanted condition, Fig. 2d–f). Comparison of the VF edge in the rest condition, highlighted in blue, with the VF edge in the implanted condition, highlighted in red, demonstrates successful implant-induced medialization in the right VF.

To quantify VF medialization, we compared VFs in rest and implanted conditions at the narrowest plane in the larynx (Fig. 2g). We defined the following parameters to evaluate the VF adduction qualitatively (Fig. 2h); detailed results are listed in Table 1 for a selected sample. A₀ is the glottal area at rest condition; overall VF adduction is calculated by (A₀−A)/A₀, where A is glottal area in the implanted configuration. D₀ is initial distance between VF medial surface and the middle plane at Z, and D is distance after implant insertion. Maximum displacement is calculated by max (D₀−D) with the location denoted as Z₁/L; maximum relative displacement is calculated by max ([D₀−D]/D₀) with the ratio denoted as Z₂/L. In the sample shown in Figure 2, the overall VF adduction ratio from the implant is 39%, indicating substantial medialization of the VF toward the midline. Medialization was observed across all samples, with variations observed in medialization between subjects consistent with differences in laryngeal anatomy.

We developed a precise, custom laser-cut silicone laryngoplasty implant which is easily reproducible. Producing implants in-house and reduced cost of labor may offset the initial investment of obtaining a laser cutter as the reduced price point represents significant savings compared to hand-carved implants or commercially available preformed implants ranging from $38 to $228 [12, 13]. The implants were designed with Tinkercad, a free and easy to use web app designed with the public in mind.

Limitations of this approach include laser cutter access and need to sterilize implants after cutting and prior to insertion. Subjects were randomized to shape/size of implant rather than using an imaging/expert-guided selection, which would likely result in improved medialization. Further, we do not evaluate phonatory parameters in this study.

This simple technology has the potential to create a library of implant shapes and sizes that can be produced within seconds at minimal cost, increasing the customizability of implants while reducing the time-cost of creating them. Laser cutting of implants is readily scalable and may present a future opportunity for clinical investigation. For our future work, we plan to produce custom laser-cut implants informed by volumetric and structural data from MRI of the larynx to study how precise implant dimensions affect VF vibration.

We would like to thank Ye Chen and Yi Song for their contributions to computational modeling, Zach Zimmerman and Alysha Slater for their assistance with surgical procedures, and William Hinson for his professional guidance and technical assistance in laser cutting. Figures were created with BioRender.com.

Animal use in this study was reviewed, approved, and monitored by the University of Pittsburgh’s Institutional Animal Care and Use Committee (approval #21120467).

The authors have no conflicts of interest to declare.

The research reported in this study was supported by the National Institute on Deafness and Communication Disorders of the National Institute of Health under Award Number 5R01DC016236 and the National Center for Advancing Translational Sciences of the National Institutes of Health under Award Number TL1TR001858. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

The project was conceived by B.R. and H.L. The methods were conceived by A.W., L.S., and G.G. A.W. and L.S. designed and implemented the methods and performed experimental data collection and analysis. Computational data collection was performed by Z.L. and H.L. A.W., L.S., Z.L., G.G., B.R., and H.L. contributed to the writing and critical revision of the manuscript. All authors have read and agreed to the published version of the manuscript.

The data presented in this study are available on request from the corresponding author. Final datasets will be publicly available when complete, per the resource sharing plan described in NIDCD award #5R01DC016236.