Enhanced construct stability with a locking plate system integrated into 3D-printed titanium implants for bone defect reconstruction: a biomechanical comparison of locking versus nonlocking screw fixation

Abstract

Background

Fixation stability is critical for successful osseointegration of 3D-printed implants in segmental bone defect reconstruction. Although locking plate systems offer theoretical advantages over nonlocking systems, particularly in osteoporotic or high-load-bearing bones, their application in 3D-printed implants remains limited owing to manufacturing challenges. We aimed to evaluate the mechanical performance of locking versus nonlocking fixation in 3D-printed titanium implants. We hypothesized that locking fixation provides superior construct stability and fatigue resistance.

Methods

Standardized synthetic bone cylinders were assembled using custom-designed 3D-printed titanium implants featuring integrated plate systems compatible with locking and nonlocking screws. Constructs were divided into two groups based on fixation method: nonlocking (n = 11) and locking (n = 11). Mechanical performance was assessed through static three-point bending (n = 6 per group) and cyclic torsional fatigue (n = 5 per group) tests. Key outcome measures included yield load, failure mode, torsional stiffness over time, and survival rate under fatigue loading.

Results

Locked constructs exhibited a significantly higher yield load under bending than nonlocked constructs (p = 0.003), indicating better resistance to irreversible deformation. Failure analysis revealed stress concentration at the innermost screw hole in the nonlocked constructs. In contrast, stress was more evenly distributed in the locked constructs, leading to bone cylinder fractures at the screw farthest from the defect. Under cyclic torsional loading, all the locked constructs survived the testing protocol, maintaining a better initial stiffness, whereas the nonlocked constructs showed progressive stiffness degradation with a 60% failure rate.

Conclusions

Locking fixation improves construct stability and fatigue resistance in 3D-printed titanium implants for bone defect reconstruction, supporting its integration into additive-manufactured implants. Future in vivo animal studies, incorporating patient-specific designs and complex loading conditions are required to validate the clinical utility of this approach.

Introduction

Repairing large bone defects, whether caused by trauma, infections, or tumor resection, remains a significant challenge in both human and veterinary orthopedics [1–3]. Traditional treatment options, including bone grafting, vascularized bone transplantation, and distraction osteogenesis, are commonly used to address these defects. However, these methods have several limitations such as the need for multistage surgeries, limited availability of bone material, delayed osseous healing, restricted early limb function, and high complication rates [1, 4, 5]. These challenges underscore the need for innovative and effective alternatives for treating segmental bone defects.

A promising alternative is the use of three-dimensional (3D)-printed patient-specific implants (PSIs). With advancements in 3D printing technology, PSIs are fabricated with precise shapes and sizes tailored towards defects. Medical-grade metals, such as the Ti6Al4V titanium alloy, offer excellent biocompatibility and mechanical strength, making them ideal structural scaffolds for repairing bone defects [3, 5]. Additionally, implant stiffness can be adjusted to prevent stress shielding by utilizing porous structures, and the porous structure promotes osseointegration through bone ingrowth within pore spaces in the implant [6, 7].

Securing an implant to remaining bone is an important factor in the design and fabrication of 3D-PSIs [8]. Fixation must provide sufficient primary stability to ensure that the defect-filling component remains in its intended position, which is a prerequisite for long-term stability with bone-implant fusion via successful osseointegration [9]. Recently, bone plate systems integrated into 3D-PSIs as a single unit have been proposed in both human and veterinary orthopaedics [3, 10–15]. These structures allow simplified surgical procedures through single-step implantation, eliminating the need for additional pre-contoured standard plates. Furthermore, the plates are tailored according to the specific configuration of the remaining bone, thereby facilitating bone fragment reduction and improving the accuracy of implant positioning.

Similar to conventional bone plates, 3D-PSIs with integrated plate systems can employ either nonlocking or locking screw fixation. Nonlocking systems rely on the compressive force generated between the plate and bone through screw torque [16]. However, this stability can be compromised in poor-quality bone or under cyclic loading, because nonlocked screws are prone to gradual loosening over time [17]. In contrast, locking systems create a fixed-angle construct through mechanical interlocking between the screw head and plate, thereby reducing the risk of screw loosening and minimizing dependence on bone quality. Locked constructs function as single-beam constructs, offering greater overall rigidity [18]. Unlike nonlocked constructs, where screws may fail sequentially, locked constructs require simultaneous failure of all screws and are thus considered more resistant to failure [17]. These characteristics render locking fixation particularly beneficial in osteoporotic and high-load-bearing areas. However, potential drawbacks such as cold welding between the screw head and plate and stress shielding of the near cortex should also be considered [17].

Although previous studies using either nonlocking [3, 13–15] or locking [10–12] plate systems integrated into 3D-PSIs have reported favorable clinical outcomes such as long-term implant survival, functional limb preservation, and minimal complication rates, to date, no study has directly compared their mechanical performance under identical implant conditions. In this study, we incorporated a locking plate system into 3D-printed titanium implant that can accommodate either locking or nonlocking screws. The implant and the bone model were designed to reflect the size of a canine segmental bone defect. To minimize confounding factors, a cylindrical bone model of uniform geometry was used. We aimed to directly compare the mechanical behavior of bone-implant constructs fixed with locking or nonlocking screws. We hypothesized that locking fixation would provide a biomechanical advantage over nonlocking fixation, which is consistent with the findings of previous studies on conventional fracture models.

Materials and methods

Construct assembly

A short, fourth-generation, fiber-filled, epoxy hollow cylinder bone model (SKU 3403-24, SawBones, WA, USA) with a length of 150 mm, a wall thickness of 2.5 mm, and an outer diameter of 10 mm was used as a bone surrogate. The cylinders were cut perpendicular to their axis to a length of 50 mm, and two segments were axially aligned, leaving a 25 mm gap at a 90° angle between them to simulate a critical-sized segmental long bone defect in a canine model [19].

The titanium implant was designed using computer-aided design software program (Rhinoceros 5.0; McNeel, WA, USA, Magics; Materialize, Leuven, Belgium). The implant comprised two main components: a shaft placed between the two segments to replace the 25 mm defect, and a fixation plate extending outward from the shaft to secure the cylinder-implant construct (Fig. 1A). The shaft included an internal mesh structure with 750 μm pores, strut thickness of 300 μm, a porosity of 75%, and a Dode design to simulate the porous PSI. The plate thickness was set to 2.0 mm, covering 70% of the length of the cylindrical length on both sides. Four threaded holes identical to those in the variable-angle locking plate system (ARIX; Jeil Medical Co., Seoul, South Korea) were included on each side of the plate, for eight holes (Fig. 1B). These holes were designed to accommodate 1.5/2.0 mm ARIX locking or cortical screws. The implants were 3D-printed in a single-step process using Ti6Al4V with a Selective Laser Melting printer (MetalSys150; Winforsys Co., Gyeonggi, Korea).

Fig. 1.

Construct assembly. A Design of the titanium implant. B 3D model of locking threads. C Implants were secured with 2.0 mm cortical screws in the nonlocking group (left) and 2.0 mm locking screws in the locking group (right)

Two groups of implant-bone constructs were assembled. In the nonlocking group (n = 11), implants were secured to the bone-substitute segments on both sides using four 2.0 mm ARIX cortical screws. In the locking group (n = 11), the implants were secured with 2.0 mm ARIX locking screws (Fig. 1C). The screws were tightened to 0.4 Nm. Six samples from each group were prepared for a single-cycle bending test, and the remaining five samples from each group were prepared for a cyclic torsion test.

Static bending test

Six constructs from each group were subjected to a single-cycle three-point bending failure test. Destructive static testing was conducted using an electrodynamic testing machine (Acumen; MTS, MN, USA) with a maximum load capacity of 25 kN. The constructs were positioned with the fixation plate on the tension side and supported by two rollers spaced 110 mm apart (Fig. 2A). A central load applicator roller applied the load at a constant compressive displacement rate of 5 mm/min, generating a load–displacement curve.

Fig. 2.

Set-up configuration. Photograph of the mechanical setup designed for (A) the single- cycle bending and (B) cyclic torsion tests. A pair of custom-made concentric clamps was attached to each end of the testing construct

The bone-implant constructs were evaluated based on their yield load (N), maximum load (N), bending strength (MPa) and failure mode. The yield load is defined as the point at which the implant begins to deform permanently, whereas the maximum load corresponds to the highest load applied before the structure fails. Bending strength was calculated according to the ASTM D790 standard [11]. Failure mode was categorized based on the observed damage, including permanent deformation of the implant, screw pullout, or bone cylinder fracture.

Torsional fatigue test

Five constructs from each group were subjected to cyclic torsion fatigue testing using a servohydraulic biaxial testing machine (Bionix; MTS, MN, USA). Each end of the construct was rigidly fixed to the machine using custom-designed concentric clamps to ensure a vertical orientation with axial rotation through the center of the shaft (Fig. 2B). Torque was applied up to a peak of ± 5 Nm in displacement control with a sinusoidal waveform, inducing a rotational angle of 0 to + 0.218 rad at a frequency of 4 Hz. Catastrophic failure (endpoint) was defined as complete fracture of the implant or bone cylinder, or complete separation of the implant-bone construct due to screw pull-out or fracture. To condition the constructs, three quasi-static load-unload cycles were performed at the start of the test. Subsequently, a quasi-static load-unload test was conducted at an angle of 0 to + 0.218 rad with a speed of 0.02 rad/s, repeated every 1,000 cycles. Testing was terminated after 30,000 cycles if no failure occurred.

Data were recorded at the start, after 25 cycles, and at every 1000-cycle interval throughout the test. Torque-angular displacement curves were generated to assess construct performance. The performance of each construct was assessed based on torsional stiffness (Nm/rad), number of cycles to failure, and failure mode. Torsional stiffness for each quasi-static load was determined from the slope of the linear portion of the torque-angular displacement curve, with the initial torsional stiffness defined from the final conditioning cycle at the beginning of the test. Additionally, percentage of the torsional stiffness relative to the initial value was calculated at the mid-point (15,000 cycles) and final cycle (30,000 cycles) of the fatigue test to evaluate the extent of stiffness degradation at key time points.

Statistical analysis

Statistical analyses were performed using GraphPad Prism 10.4.0 for Windows (GraphPad Software, CA, USA). Normality of data distribution was evaluated using the Shapiro–Wilk test. Because the data met the assumption of normality, yield load, maximum load, and bending strength (from the static test) between the groups were compared using an independent-samples t-test with the Welch's correction. Torsional stiffness data (from the fatigue test) were not normally distributed; therefore, the Wilcoxon rank-sum test was used to compare the initial torsional stiffness between the groups, although the Friedman test followed by the Bonferroni-Dunn post-hoc test were applied for within-group comparison of torsional stiffness over time. Statistical significance was set at p < 0.05 for all tests.

Results

Static bending test

The locking group had a significantly higher yield load than the nonlocking group (p = 0.003) (Table 1). The average values of the maximum load and bending strength were higher in the locking group, but no statistically significant differences were observed. The failure modes differed between the nonlocking and locking groups (Fig. 3A). In the nonlocking group, five failures occurred because of plate breakage at the screw hole nearest from the implant shaft, whereas one failure occurred because of bone cylinder breakage at the outermost screw hole area. In the locking group, one failure was due to plate breakage at the innermost screw hole area, and five failures resulted from bone cylinder breakage. Of these, two failures occurred at the outermost screw hole area, and three failures occurred at the second outermost screw hole area, which were accompanied by fracture of the outermost screw head.

Table 1.

Summary of static bending test data

Variable	NL (Mean ± SD)	L (Mean ± SD)	P-value
Yield load (N)	683.25 ± 53.80	801.78 ± 48.00	0.003
Maximum load (N)	1117.50 ± 121.00	1169.82 ± 31.53	0.35
Bending strength (MPa)	184.39 ± 19.96	193.02 ± 5.20	0.35

Fig. 3.

Modes of failure. A Construsts after static bending test. Typical failure modes of the nonlocking (left) and locking (right) groups. The arrow indicates a broken screw head that remained seated in the screw hole, maintaining the locked configuration. B Construct after the fatigue torsion test. Failure mode of the nonlocking group. Breakage of the screws occurred on the threaded body, which were secured in the bone

Torsional fatigue test

All the specimens in the locking group (5 of 5, 100%) survived the testing, whereas 3 of 5 specimens (60%) in the nonlocking group failed during the cyclic loading. The failure mode was consistent across all failed constructs in the nonlocking group, in which four screws in the lower grip broke in sequence, resulting in complete separation of the implant from the bone cylinder. (Fig. 3B). Breakage of the screws occurred on the threaded body, secured to the bone.

The initial torsional stiffness was higher in the nonlocking group compared with that in the locking group (median 13.7 Nm/rad versus 9.6 Nm/rad), but this difference was not statistically significant (p > 0.05). Evolution of average stiffness over time differed between the two constructs (Fig. 4). The nonlocked constructs showed several abrupt decreases in stiffness, with a statistically significant decrease from the initial stiffness occurring at 12,000 cycles. In contrast, the locked constructs exhibited only one initial drop, followed by a gradual decline, with a significant reduction from the baseline observed in 23,000 cycles. Consequently, the locking group maintained a higher percentage of its initial stiffness by the end of the test. The percentage changes in torsional stiffness relative to the initial value were 67.8% and 69.3% for the nonlocking and locking groups, respectively, at 15,000 cycles, and 4.4% and 63.0% at the end of the test, respectively, indicating a substantially greater loss of stiffness over time in the nonlocked constructs.

Fig. 4.

Evolution of torsional stiffness over time. Evolution of torsional stiffness over time for the nonlocked (round) and locked (square) construct

Discussion

Our study demonstrated that locking screw fixation enhanced the construct stability of 3D-printed titanium implants for canine segmental bone defect reconstruction compared with nonlocking fixation. Specifically, the locked constructs exhibited a higher yield load under bending and better maintenance of torsional stiffness under cyclic loading. These findings align with the theoretical advantages of locking plate systems, which provide angular and axial stability through a threaded interface between the screw heads and plate body, enabling the construct to behave as a single beam [20]. Similarly, previous studies have compared the mechanical stability of locked and nonlocked fixation in fracture models and reported that locked constructs generally outperform nonlocked constructs, particularly in osteoporotic bone models [20–23]. Although variations in bone quality, loading conditions, and experimental settings have led to inconsistent outcomes across studies, the biomechanical advantage of locking fixation has been consistently observed.

Despite these advantages, the application of locking fixation to 3D-printed implants remains at an early stage of development. A primary challenge is the technical difficulty in fabricating reliable locking threads through additive manufacturing, owing to their small and intricate geometry, which poses challenges both in absolute terms and in relation to the overall part structure [24]. Owing to the fine thread geometry, post-processing steps such as computer numerical control machining or tapping are often necessary, increasing both production time and cost [25]. To address these limitations, we adopted the ARIX locking system as a reference. This is a commercially available variable-angle locking plate system characterized by its simplified thread geometry with internal threads confined to the lower part of the plate hole, making it well-suited for additive manufacturing. This design not only facilitates direct 3D printing without additional machining but also provides the potential for screw angulation up to 15°, which, although not utilized in this study, may offer clinical advantages in accommodating various anatomical situations.

In the static bending tests, the locked constructs demonstrated a significantly higher yield load than the nonlocked constructs, indicating that locking fixation enables the construct to withstand greater forces before irreversible deformation occurs. This difference could be attributed to the distinct load-bearing mechanisms of the two fixation systems. In nonlocked constructs, stress initially concentrates on the tightest screw, causing early loosening and subsequent transfer of load to neighboring screws. In contrast, locked constructs act as a single beam, with all locked screws collectively resisting pull-out forces, thereby withstanding mechanical loads more effectively [18]. The observed failure patterns further support these findings; plate fractures in the nonlocked group occurred at the innermost screw hole, where stress was most concentrated during bending, whereas in the locked group, stress was distributed more evenly, leading to bone cylinder fractures at the screw farthest from the defect. These patterns are consistent with those of previous studies on fracture models fixed with nonlocking or locking screws [26, 27].

Interestingly, although the locked group demonstrated a higher yield load during bending, it is worth noting that both groups exhibited clinically acceptable strength. For example, Bertram et al. [28] reported that peak vertical forces of approximately 76–107% of the body weight are transmitted through the hindlimbs of normal Labradors and Greyhounds during trotting. Hutcheson et al. [27] later applied these data to evaluate the construct performance, providing a clinically relevant frame of reference for bending tests. For a 40 kg dog, this corresponds to approximately 297–419 N of force. In comparison, plastic deformation occurred at 683 N in the nonlocked constructs and 802 N in the locked constructs in the present study, representing approximately 1.6–2.3 (nonlocked) and 1.9–2.7 (locked) times the estimated physiological load. These results suggested that the additively manufactured plate holes in our study provided reliable performance, demonstrating clinically acceptable strengths for both fixation methods. Although either nonlocking or locking fixation may be selected based on clinical scenario, locking fixation provides additional mechanical stability.

To further evaluate the construct stability under conditions simulating repetitive physiological loading, cyclic torsional testing was conducted, as torsion represents a critical force during ambulation and is a common failure mode for plate and screw fixation in dogs [29]. The locking group demonstrated superior maintenance of initial stiffness and achieved complete survival throughout the test duration. In contrast, in the nonlocking group, repetitive loading led to screw loosening with progressive stiffness loss, and the lever-arm effect caused sequential screw fracture in 60% of the specimens. Interestingly, the initial torsional stiffness was higher in the nonlocked group, although this difference was not statistically significant. Given that our tests employed relatively strong synthetic epoxy bone models, the nonlocked constructs likely benefited from greater plate-to-bone friction, partially masking the advantages of locking fixation [30]. In support of this, Ricci et al. [17] reported that locking constructs outperformed nonlocking constructs under cyclic axial loading in osteoporotic but not non-osteoporotic bones, highlighting the effect of bone quality. These findings suggest that the advantages of locking fixation observed in the present study may be further amplified in osteoporotic or pathological bone conditions, and further investigations may enhance our understanding of its clinical application.

Although this study successfully demonstrated the feasibility of integrating a locking plate system into 3D-PSIs and highlighted its clinical relevance by showing superior construct stability compared with nonlocking systems, several limitations should be acknowledged. We used synthetic bone models rather than cadaveric bones, which limited the direct clinical extrapolation of our findings. However, this approach allowed us to minimize the variability in bone quality and morphology, thereby isolating the mechanical effects of the fixation method. Additionally, although we focused on bending and torsional loading—given that they collectively account for approximately 95% of physiological loading [31],—additional loading conditions, including axial compression, shear forces, and combined multiaxial stresses, should be considered in future studies to provide a more comprehensive understanding of the effects of fixation methods on implant-bone constructs. Furthermore, the relatively small sample size for each mechanical test (n = 5 per group for torsion) may have limited the ability to detect subtle differences between groups, potentially resulting in Type II errors. Future studies with larger sample sizes are warranted to confirm these findings. Finally, although only purely locked and nonlocked constructs were tested, combining a compression screw near the osteotomy with locking screws for overall stability may represent a practical alternative. Such a hybrid approach could improve bone–implant contact while retaining the rigidity of locked fixation.

Despite these limitations, our study successfully demonstrated the mechanical feasibility of integrating a locking plate system into 3D-printed PSIs and provided valuable insights into their biomechanical performance. The in vitro tests confirmed the superior primary stability of locking fixation, although it remains to be verified whether this mechanical advantage directly translates to improved biological outcomes such as osseointegration. Additional in vivo investigations in canine models, particularly under compromised bone conditions or more complex loading environments, are warranted to validate these findings and assess their biological relevance. Importantly, the present work establishes a preclinical foundation in a canine segmental bone defect model, which may guide the design and selection of patient-specific implants in clinical veterinary practice and provide insights for potential translational applications in human orthopedic procedures.

Conclusions

This study confirmed that the choice of fixation method plays a pivotal role in the mechanical performance of 3D- PSIs. Locking fixation provided superior construct stability and fatigue resistance compared to nonlocking fixation, thereby supporting our hypothesis. Given the importance of primary stability in facilitating osseointegration, especially in challenging clinical scenarios such as osteoporotic or high load-bearing bones, our findings underscore the clinical relevance of integrating locking systems into 3D-printed implants for segmental bone defect reconstruction.

Authors’ contributions

KK carried out the experiments, analyzed and interpreted the data, and drafted the manuscript. SY was involved in the design and manufacturing of the implant. SK participated in the conceptualization and supervised the analytic process and provided critical feedback on the manuscript. BK contributed to the study design, supervised the overall project, and assisted with funding acquisition and manuscript revision. All authors discussed the results, contributed to the final manuscript, and approved the submitted version.

Funding

This research was supported by the Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education (RS-2024-00407214) and by another NRF grant funded by the Korea government (MSIT) (No. 2023R1A2C1003001).

Data availability

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethical approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.