Degradation and Fatigue Behavior of 3D-Printed Bioresorbable Tracheal Splints

Abstract

Introduction:

Severe infantile tracheobronchomalacia (TBM) is often treated with invasive surgery and fixed size implants to support the trachea during respiration. A novel 3D-printed extra-luminal splint has been developed as a flexible and bioresorbable alternative. Therefore, the goal of the present study was to use an in vitro breathing simulator model to comprehensively evaluate the structural stiffness and failure modes of two sizes of a novel bioresorbable 3D-printed splint design under a range of physiological degradation conditions.

Methods:

Two thicknesses, 2mm and 3mm, of a novel 3D-printed bioresorbable splint were evaluated under two different degradation conditions, phosphate-buffered saline (PBS) and sodium hydroxide (NaOH). The splints were subjected to simulated breathing loading, involving a cyclic opening and closing of the splint by 2mm, for a targeted duration of 7.5 to 30 million cycles. A separate new set of splints were statically soaked in their respective degradation condition for a comparative analysis of the effects of cyclic loading by the degradation medium. After successful simulated breathing or static soaking, non-destructive tensile and compressive strengths were evaluated, and overall stiffness was calculated from destructive tensile testing.

Results:

The present study indicates that the splints were more significantly degraded under simulated breathing conditions than under soaking. Cyclic simulated breathing specimens failed far earlier than the intended duration of loading. Over time, both 2mm and 3mm splints became increasingly more flexible when subjected to the static degradation conditions. Interestingly, there was little difference in the compressive and tensile strengths of the 2mm and 3mm thickness splints.

Discussion:

The bioresorbable nature of PCL offers a valuable advantage as it eliminates the need for splint removal surgery and increases device flexibility over time with degradation. This increased flexibility is crucial because it allows for uninhibited growth and development of the infant’s trachea over the intended use period of two years. The results of this study confirm that the splints were able to withstand tensile forces to prevent tracheal collapse. This study further supports the successful use of 3D-printed splints in the treatment of infantile TBM.

Introduction

Pediatric tracheobronchomalacia (TBM) is a rare condition that can result in the collapse of the respiratory airways, potentially leading to fatal cardiopulmonary events.^1–5 Aggressive early intervention during infancy is crucial to prevent respiratory failure and pulmonary failures.⁶ Current treatment options include cardiovascular surgery to correct the anatomy and decompress the trachea, tracheostomy with long-term artificial ventilation, and intra-luminal airway stents. However, these treatments are associated with severe complications, such as granulation tissue growth and mechanical failures, resulting in a mortality rate of 7-8% in infants.^7–10 Furthermore, existing stent treatments have high rates of stent fractures, migration, and only short-term symptom relief.^11–13

Current stent treatments typically utilize fixed sizes which may obstruct diameter expansion.¹⁴ Subsequently, these stents may fail due to an inability to expand as the trachea grows or may have too much mobility to support tracheal growth, ultimately obstructing natural resolution of TBM.¹⁴ In addition, stents may erode through the airway, migrate and cause granulation tissue formation inside the airway.^15,16 To address these complications, a novel 3D-printed, flexible, extra-luminal splint was designed as a bioresorbable alternative for severe pediatric TBM cases. Polycaprolactone (PCL) was selected due to its flexibility, allowing for tracheal expansion. Further, the aliphatic ester linkages in PCL are susceptible to hydrolysis which allows the resorption of PCL over a period of approximately 24-36 months. This aligns with the identified period of new tissue growth in child development, making PCL an ideal material for this application.^3,17,18

During natural breathing patterns, these splints are subject to cyclic expansion and elastic deformation making them prone to fatigue failure and plastic deformation. Therefore, the goal of the present study was to evaluate the structural stiffness and failure modes of two sizes of a novel bioresorbable 3D-printed splint design using an in vitro breathing simulator model.

Methods

Implants were manufactured in two wall thicknesses, 2mm and 3mm, with undulating, bellowed wall thickness of 15% of maximum wall thickness for the 2mm designs and 45% of maximum wall thickness for the 3mm designs. These designs were assessed under a variety of conditions (Figure 1a–c). Two degradation conditions, (1) immersed in phosphate-buffered saline solution (PBS) to model real-time degradation and (2) immersed in sodium hydroxide (NaOH), to simulate accelerated degradation, and two loading conditions, (1) simulated breathing (expanding and contracting the splint) and (2) static soak only, were assessed at different lengths of time (Figure 2). A set of six implants (N=6) for each wall thickness were evaluated for 13 experimental conditions, comprised of different combinations of the degradation and loading conditions and time (N = 150 total specimens) (Figure 2). Each specimen was tested under non-destructive tensile and compressive testing prior to degradation and/or cyclic loading. Non-destructive testing was repeated after degradation and/or cyclic loading, followed by destructive tensile load-to-failure testing.

Figure 1.

Tracheal Splints and Testing Apparatus. (a) – (c) Depiction of 3D printed tracheal splints (2mm on left and 3mm on right) (d) Tensile testing apparatus for tracheal splints (e) Compressive testing apparatus for tracheal splints (f) Simulated breathing loading setup (Movement of splints denoted by red arrows)

Figure 2.

Outline of the Six Loading and Degradation Conditions.

Non-Destructive Static Stiffness Testing

The stiffness of each specimen was characterized prior to degradation and/or cyclic loading. A custom loading apparatus was designed to conform to the inner contours of the splint for tensile testing using a biaxial servo-hydraulic load frame (858 Mini Bionix; MTS Systems, Eden Prairie, MN). Rather than applying a point force to the edges of the splint, this apparatus was intended to distribute the forces within the inner walls, simulating the expansion of the trachea during breathing (Figure 1d). Once the specimens were pre-loaded, distraction was applied to produce a 2mm gap at the open edge of the splint. To evaluate the compressive stiffness, an apparatus was designed to conform to the outer contours of the splint (Figure 1e). Specimens were again pre-loaded, and compression was applied to produce 2mm of gap-closing at the open end of the splint.

Soaked (Non-Loaded)

Specimens were immersed in (1) PBS or (2) NaOH in individually labeled, sterile, and airtight containers. Four complete sets of 2mm and 3mm specimens (N=48) were immersed in PBS solution for 6-month intervals between 6-24 months. Two complete sets of 2mm and 3mm specimens (N=24) were immersed in NaOH for 2 or 4 months. Interim characterization, including non-destructive stiffness testing and visual inspection, was conducted every 3 months for the PBS-soaked specimens, and every month for the NaOH-soaked specimens. After the completion of degradation durations, specimens were loaded in non-destructive tensile and compression testing, followed by the tensile-load-to-failure protocol.

In order to assess the effect of degradation, the change in force to create a 2mm gap as a result of degradation was calculated for each specimen and then averaged for each group. All values were normalized by subtracting the initial force of the specific specimen from the force required for either tension or compression after the degradation and loading conditions were applied.

Cyclic Loading (Simulated Breathing)

A custom apparatus that conformed to the inner contours of the splints was developed to cyclically load implants and simulate the physiological conditions induced by the opening and closing of the airway (Figure 1f). The apparatus was mounted in an MTS multi-station Bionix Wear Simulator (MTS Systems, Eden Prairie, MN). Periodical monitoring of each specimen, including weekly fracture checks and daily fluid refills, was performed to ensure the integrity for the duration of the test. The testing protocol was based on ASTM F3036-13 Standard Guide for Testing Absorbable Stents. All testing was conducted at a rate of 1.0Hz. Four complete sets of 2mm and 3mm specimens (N=48) were immersed in PBS solution and intended to be tested for durations of 7.5M, 15M, 22.5M, and 30M cycles at 1.0Hz. One complete set of 2mm and 3mm specimens (N=12) was immersed in NaOH and intended to be tested for 5M cycles.

Tensile Load-to-Failure Protocol

Three specimens from each testing condition group were loaded to failure under tension after completion of the respective degradation conditions and nondestructive tensile and compressive testing. Failure was defined as any severe, permanent mechanical damage seen in the splint, such as fractures or complete splint breakage.

Results

Overview

All 150 specimens successfully completed pre-degradation non-destructive tensile and compressive testing following degradation conditions. Further, all soak-only specimens (both PBS and NaOH) were viable for post-degradation nondestructive tensile and compressive testing, and destructive tensile testing. However, in the simulated breathing groups (both PBS and NaOH), no specimens remained intact for the intended duration of cyclic loading and, therefore, only six specimens (N=3, 2mm and N=3, 3mm) had interim measurements for analysis. Failure patterns were noted with most fractures observed at the hole interface, as described for each group below.

Pre-degradation Non-Destructive Stiffness Testing (All 4 groups)

Pre-degradation, non-destructive tensile testing was done on all specimens at time zero months to develop a baseline distraction force. The average tensile force at 2mm of distraction was 5.8N ± 0.6 for the 2mm specimens and 10.5N ± 0.9 for the 3mm specimens. This difference was statistically significant (P < 0.01). The average compressive force at 2mm of compression was 5.6N ± 0.8 for the 2mm specimens and 3.2N ± 2.2 for the 3mm specimens. This difference was not statistically significant (P > 0.05).

PBS Soak Only, Post-Degradation

Non-Destructive

Specimens that were soaked in PBS solution were evaluated at 6-month intervals from 6 to 24 months. With original data, after 6 months, 12 months, and 18 months of PBS soak-only, the average tensile force for non-destructive testing was not statistically significant between the 2mm and 3mm specimens. However, after 24 months of PBS soak, the average tensile force to distract the splint by 2mm was significantly different between the 2mm and 3mm specimens (P < 0.05). Interestingly, following normalization of data, using the pre-degradation non-destructive tensile force to distract by 2mm, significance was diminished and the difference between 2mm and 3mm was only significant for 6 months of soak, but not significant at any other soak-only length of time (P > 0.05) (Figure 3).

Figure 3.

Tension and compression testing results for PBS soak specimens in static, non-destructive testing. Each specimen was tested after degradation and normalized according to the pre-degradation value for that specimen. Specifically, the values shown are obtained by (PBS-Soak Force (N)) – (Original Dry Force (N)). Accordingly, a positive value indicates that the PBS-soaked force was higher than the original dry force. Significance of P < 0.05 is denoted by a *.

At all soak-only time points (6, 12, 18, and 24 months), there was a statistically significant difference in the compressive force required to compress the splint by 2mm (P < 0.05). However, when normalizing these values to baseline, the significance diminished following 6 months of soaking (P > 0.05). In contrast, the significance between the change in compressive force remained following 12 months, 18 months, and 24 months of soak-only (P < 0.01) (Figure 3).

Destructive

Average stiffness values were calculated for each specimen following tensile load to failure. At all PBS soak-only time points, the difference in stiffness of 2mm and 3mm specimens was not statistically significant (P > 0.05) (Figure 4).

Figure 4.

Stiffness results calculated from tensile load-to-failure testing. Specimen stiffness was calculated using the slope of the axial displacement (mm) versus force (N) graph when specimens were distracted in tension until failure. Significance of P < 0.05 is denoted by a *.

NaOH Soak Only, Post-Degradation

Non-Destructive

After soaking in NaOH for 2 months, significant differences were seen between the 2mm and 3mm average tensile forces at 2mm of distraction (P < 0.01). However, after 4 months of NaOH soak-only, there was no statistical difference between the 2mm and 3mm average tensile force to distract the splint by 2mm (P >0.05). When comparing the normalized values to the baseline pre-degradation non-destructive tensile testing, the statistical significance changed. Specifically, the change in tensile force from baseline to distract the specimen by 2mm was no longer significant following 2 months of soak-only (P > 0.05) (Figure 5). In contrast, following 4 months of soak only, the change in tensile force from baseline became significant between 2mm and 3mm specimens (P < 0.01) (Figure 5).

Figure 5.

Normalized tension and compression testing results for NaOH soak specimens in static, non-destructive testing. Each specimen was tested after degradation and normalized according to the pre-degradation value for that specimen. Specifically, the values shown are obtained by (PBS-Soak Force (N)) – (Original Dry Force (N)). Accordingly, a positive value indicates that the PBS-soaked force was higher than the original dry force. Significance of P < 0.05 is denoted by a *.

The average compressive force to compress the splint gap opening by 2mm was not significantly different between the 2mm or 3mm specimens following both 2 months and 4 months of NaOH soak only (P > 0.05). However, when normalized to baseline values, the change in compressive force from baseline for both 2 months and 4 months of soak only demonstrated significant differences between the 2mm and 3mm specimens (Figure 5).

Destructive

Average stiffness values were calculated for each specimen following tensile load to failure. Following 2 months of NaOH soaking, the stiffness of the 3mm specimens was significantly larger than the stiffness of the 2mm specimens (P < 0.05). However, after 4 months of NaOH degradation, there was no difference between the stiffness values of the 2mm and 3mm specimens following tensile load to failure (Figure 4).

PBS Soak Simulated Breathing

Most specimens cyclically loaded in PBS solution successfully completed one million cycles of simulated breathing. The 2mm specimens failed at an average of 1,068,619 ± 502,210 cycles, while the 3mm specimens, failed slightly quicker, on average 1,036,029 ± 508,815, though this difference was not statistically significant (P = 0.90) (Figure 6). Three 2mm and three 3mm specimens soaked in PBS achieved 1.5 million cycles of testing without failure and were able to withstand non-destructive tensile and compressive testing for the first interim measurement point. Non-destructive testing of these specimens indicated that, as expected, tensile force to distract the splint by 2mm significantly decreased following 1.5M cycles of loading when compared to baseline conditions (P < 0.01). As expected, 2mm and 3mm specimens in the PBS soak simulated breathing group were significantly weaker in both tension and compression than the PBS soak-only specimens at 3 months (P < 0.01) (Figure 7). Further, no significant differences were reported between average tensile or compressive forces of the 2mm and 3mm specimens following 3 months of simulated breathing (P > 0.07).

Figure 6.

Cycles before failure. Any crack or separation observed denoted failure. The 3mm specimens with NaOH soak have no standard deviation as failure of all specimens was noted at the same cycle. Significance of P < 0.05 is denoted by a *.

Figure 7.

Cycles before failure. Any crack or separation observed denoted failure. The 3mm specimens with NaOH soak have no standard deviation as failure of all specimens was noted at the same cycle. Significance of P < 0.05 is denoted by a *.

After non-destructive interim testing, all six specimens were returned to their chambers and resumed cyclic testing; however, all specimens failed prior to the next intended interim measurement and no further non-destructive measurements could be obtained. For specimens that failed during cyclic simulated breathing, the observed failures were all fatigue fractures with cracks that propagated from the design’s suture hole (Figure 8d).

Figure 8.

Tracheal Splint Fracture Types. (a) – (c) common fracture or deformity types that constituted a splint failure (d) complete fracture seen in many of the shorter soak specimens due to the brittleness of the device (e) hairline fracture noted in the longer soaked specimens due to increased flexibility prior to failure.

NaOH Soak, Simulated Breathing

As expected, specimens soaked in NaOH failed earlier, due to accelerated degradation, than specimens tested in PBS. The 2mm specimens failed at an average of 45,508 ± 33,430 cycles, while the 3mm specimens withstood slightly more cycles of loading, 67,087 ± 0 (P = 0.15) (Figure 6). As daily fracture tests were done, all specimens were detected to have failed at the same time point, thus there is no standard deviations for the 3mm specimens. Further, as no specimen reached the first interim measurement time point, no non-destructive tensile or compressive testing was possible. The differences in the number of cycles to failure for NaOH-soaked specimens when compared to PBS-soaked specimens, regardless of size, were significant (P < 0.01) (Figure 6).

Mechanisms of Failure

The cyclic simulated breathing failures, for both PBS and NaOH, were all fatigue failures with cracks observed at the hole interface. For specimens in the soak-only groups, both PBS and NaOH, fracture patterns during tensile load to failure testing at early time points was consistent with the simulated breathing fractures (Figure 8a–d). However, as the period of soaking increased past 12 months, the splints exhibited heavy deformation due to increased malleability (Figure 8b & e).

Discussion

The goal of the present study was to assess the performance of novel 3D printed, bioresorbable tracheal splints as a function of degradation over time. While these splints have been clinically successful in the treatment of infantile TBM thus far, no study has biomechanically evaluated the performance of these implants.^19,20 In vivo conditions were simulated to characterize their ability to perform throughout the two years of intended use. This study successfully identified the factors that heavily influence degradation and future opportunities for improvement of these devices.

Simulated Breathing Loading Analysis

Simulated breathing was found to be considerably more detrimental to the strength of the splints than soak-only. Three months of simulated breathing dramatically decreased the strength of the splints compared to soak-only. Therefore, the degradation of these splints may be more dependent on the cyclic forces from breathing rather than the physiological conditions of the body that cause degradation.

Cycles to fail during simulated breathing varied based on the different degradation conditions. As anticipated, the specimens in the accelerated degradation conditions, NaOH, degraded faster and failed after the fewest cycles of loading. This may indicate that the simulated in vivo degradation condition may actually be more conducive to long-term cyclic loading. The PBS soak may have allowed for specimens to last longer in a simulated breathing cyclic environment and provide a necessary lubrication for these splints.

Regardless, none of the specimens that underwent simulated breathing withstood their intended duration of cyclic loading. This may indicate that either the design was not strong enough or the loading conditions were too extreme. Although breathing rate in children decreases as they age, the average number of breaths per minute for a child between the age of 0 to 24 months is about 37.5 breaths/minute.²¹ Assuming consistent breathing at this average rate, the frequency of tracheal opening and closing for an infant would be about 0.63Hz.²¹ The frequency for simulated breathing in the present study was 1Hz. Although this is a small discrepancy, this may be a factor that influenced the early failures of these devices.

Failure Type

Because the flexibility of these splints is necessary to support the expanding tracheal diameter and allow for natural regrowth of tissue, an increased malleability over time may be advantageous to the success of these splints. Splints that underwent six months of soaking in PBS quickly fractured into two fragments, similar to the original 0 month fractures, indicating the retention of the original material properties of the splints. However, as soaking increased to 24 months, the splints deformed considerably without complete fracture, indicating increased flexibility. Therefore, as the material degrades and softens due to the physiological environment, the splints have more plastic deformation, which will allow for tracheal expansion.

The flexibility of these devices seen in extended soaking conditions contributes to the potential for TBM treatment success. As the trachea expands during natural regrowth, the device’s flexibility can allow for infant tracheal growth without the potential for intra-tracheal granulation tissue buildup.¹³ The development of granulation tissue in previous TBM treatments is a common complication that results in airway obstruction. This is typically due to splint inflammation.^22,23 An extra-luminal splint may avoid the inflammation and subsequent formation of granulation tissue by not introducing intraluminal foreign bodies. Further, the natural tracheal regrowth and bioresorbable properties of these devices negate the need for removal surgery and may decrease the incidence of airway obstruction due to granulation tissue formation.

In the present study, failure of the splints following simulated breathing were typically fatigue failures occurring at the hole interface. These failure patterns mimicked those seen in the tensile failure tests for soaked specimens at early time points. Based upon these results, the hole interface is identified as the weakest point. This pattern intends to provide periodic locations for fixed suturing and to improve the flexibility of the device during tracheal growth. This study suggests the hole pattern could be optimized to improve design performance.¹⁹ Device improvements may include increasing distances between the holes, decreasing the number of holes, or staggering the pattern of holes.

Clinical Significance

Based on clinically observed changes in tracheal diameter, a 2mm distraction or compression was selected (Table 1). This represents a 13% change in diameter in either compression or tension, consistent with the physiological expansion and contraction of the trachea during breathing.²⁴ Therefore, the applied distraction was intended to replicate physiological respiratory changes in a healthy individual. Clearly, the resistance to the distraction forces applied to the splint would likely decrease over time as the splint biodegrades. Ideally, the healing of the trachea over with time would render the splint unnecessary.²⁵ However, the probability of distractions larger than 2mm is relatively small, as this model represented a worst case scenario to be fail-safe in infants.

Table 1.

Tracheal Splint Distraction Parameters. Based on a previous study data, proportion of tracheal diameter and change in diameter following inhalation, the 2.0 mm distraction parameter was determined.

Trachea		Diameter (mm)	Change in Diameter (mm)	Percent Change (%) (|Change Diameter/Diameter|)
Splint		15	2.0	13.3
Malacial (0-7 years)	Lower Limit	1.4	0.2	14.3
	Upper Limit	7.7	−0.9	11.7
Healthy (0-7 years)	Lower Limit	4.5	0.6	13.3
	Upper Limit	8.28	−1.0	12.1

Current success of these extra luminal splints and the resulting dramatic decrease in TBM symptoms has already been reported in three infants with a 2.5-year follow-up period.²⁰ Another study examined the performance of the present customized tracheal splints followed 15 patients with 29 splints at a median age of 8 months.¹⁹ The first four patients who received splint implantation exceeded the 24 month splint survival period which indicates sustained and significant clinical improvement. Additionally, no patient required splint re-operation. The results from the present study with 24 month performance testing corroborate findings from both studies, indicating the success of these splints despite the degradation and fatigue we observed.

Previous studies have reported success in the use of extra-luminal splints but have not accounted for some major complications. One study compared extra-luminal rings and intra-luminal splints for the treatment of tracheal collapse in dogs.²³ These extra-luminal rings led to paralysis of laryngeal nerves in 10-30% of dogs due to nerve damage during insertion and tracheal necrosis that resulted from disruption of the trachea’s blood supply due to compression from the extra-luminal rings.^26–30 Intra-luminal stents reported tracheal collapse in 83-89% of cases after only 1 year of use.³¹ Further, intra-luminal stents have had high reports of splint migration and formation of granulation tissue.^32,33 Studies have assessed a polytetrafluoroethylene (PTFE) extra-luminal splint for treating TBM.^34,35 However, deaths due to respiratory distress, congestive heart failure, and ventricular dysfunction were some of the causes reported in 14 of 127 patients who received these splints. Additionally, 9 patients reported complications and reoperations due to recurrent obstructions.³⁵

The device in the present study may potentially address these complications since it is intended to allow greater flexibility and decreased compressive forces on the trachea. Extra-luminal implantation of the flexible splints is intended to decrease splint migration and granulation tissue overgrowth into the splint. Furthermore, the present splints maintained shape and integrity following two years of simulated in vivo degradation.^17,36 Additionally, while increased flexibility over time enables natural growth of the trachea, force distributions found in the present study may help the device to resist tracheal collapse and negate the need for infant exposure to risky TBM treatments.

Device Thickness

In the present study, the 2mm thickness splints produced similar strength and stiffness results to the 3mm splints. These results suggest that the use of a 2mm splint has advantages, including decreased prominence leading to lower incidence of irritability. Patients with extra-luminal splinting have reported increased coughing episodes due to its proximity to the larynx.²³ In the present study, the 3mm splints, with increased material and size, were expected to be significantly stronger; however, there was no notable difference in device performance based on thickness. An additional advantage of a thinner device, such as the 2mm splint in the present study, may be that it requires a smaller tensile force to distract the device enough for sufficient tracheal opening while being splinted in the infant.

Limitations

There were several limitations in the present study. First, this is an in vitro model which cannot replicate all the physiological in vivo conditions that the splints will be subjected to. Further, this is a simplified model of breathing that applies a constant tension to distract the splint by 2mm but does not account for any irregularities. Furthermore, the rate at which the trachea regrows and how that may affect the loading conditions is unknown. Therefore, this simplified loading condition may not accurately represent the loads applied to the splints after significant tracheal regrowth. Additionally, the degradation medium used was intended to simulate the in vivo conditions of the splint, however, degradation in the body may be faster. The static testing of tension and compression required a small amount of load to be applied prior to testing and this starting point was slightly subjective. To minimize the subjectivity of the starting point, the same technician loaded each device statically. Additionally, splints were soaked at room temperature, 22°C, whereas body temperature is around 37°C which could change the degradation characteristics of the splints. Lastly, the size of the splints used in this study were larger than the sizes for infants, however, the percentage opening was comparable and suggest that the strain on the critical locations is probably similar.

Conclusion

The present study quantified the degradation properties of 3D printed, bioresorbable splints designed for the treatment of infantile TBM under simulated breathing conditions. The static soaked splints had consistently maintained mechanical properties over 24 months of degradation, suggesting the potential to provide sufficient resistance to tracheal collapse in patients. However, cyclic simulated breathing conditions resulted in early failures of these novel splints. Future studies could explore alternative hole patterns or modifications to improve the strength and durability of the splints. Overall, these findings demonstrate the potential for 3D-printed splints as a bioresorbable alternative to current treatments for TBM in infants.