The mesh structure of the talus implant was designed using periodically repeated unit cells. Numerous mesh structures have been studied and verified in the literature, demonstrating that the structure, shape, and unit cell size significantly contribute to the mechanical performance of customized implants [28, 29]. Considering the specific requirements for the talus implant, such as low weight, sufficient load-bearing capacity (approximately 100 kg), and large cell size to accommodate a bone graft, the unit cell size was set to be 10 mm.

In this study, we considered two different unit cell structures for the talus implant: (i) the “rotating cross” structure (Fig. 2), (ii) the “rhombic dodecahedron” structure (Fig. 3). The relative density of all unit cells was set to 30%.

To assess the performance of this lattice structure, a load of 100 kg (approximately 980 N) was applied, representing the weight of a patient. Considering the surface area of the talus (approximately 0.002 m²) and load distribution between both feet, the load applied to the mesh structure for talus loading was approximately 73.5 kN/m².

Simulation results indicated that for “Rotating Cross” lattice structure stress concentration occurred when the load was applied in directions other than the normal direction (Fig. 2). Stress reached up to 25 MPa at the edges of the nodes where struts are connected, while normal loading resulted in stresses less than 2.8 MPa. It means shearing or twisting (e.g., loading at 45°) may lead to a fatal outcome. For the “Rhombic Dodecahedron” lattice structure, the loading direction has minimal effect (Fig. 3).

Simulated results for the “Rotating Cross” lattice structure. The figure illustrates overall view of the 45° loaded samples (a), where stress reached to 25 MPa at the edges of nodes (b). The overall view of the samples loaded in the normal direction (c) shows that stress reaches to 2.28 MPa at the struts (d)

Simulated results for the “Rhombic Dodecahedron” lattice structure. The figure illustrated overall view of the samples loaded in the normal direction (a), where stress reached to 10.1 MPa at nodes where struts are connected (b)

To verify the lattice structure’s tolerance to mechanical loading, samples were tested in a universal testing machine. In Fig. 4 the applied force vs. strain (blue curve) and generated stress vs. strain (orange curve) are depicted for the “Rotating Cross” sample with a size of 19 × 19 × 19 mm. As Fig. 4 clearly shows the sample was compressed up to 3.5% without fracture; however, plastic deformation occurred around 0.7–1.5%, indicated by a distinct transition zone in the stress-strain curve.

Stress-strain curve and loading path for a cubic sample with the “Rotating Cross” structure. The blue curve illustrates applied force vs. strain, while orange curve shows the generated stress vs. strain

After three cycles of loading and unloading, the samples failed (Fig. 4). The failure may be attributed to the small rod dimensions during the printing process and stress concentrations at the strut connections.

In conclusion, even though the stress is well below the yield point of the Ti alloy, the sample failed to pass the fatigue test. This may be related to both poor printing processes due to small rod dimensions and stress concentrations at the strut connections. Both factors are important, hence we need to evaluate which one has the greater impact.

Figure 5 shows the SS curves and load hysteresis for a “rhombic dodecahedron” lattice (Fig. 5a) and dynamically loaded and unloaded sample response (Fig. 5b). Mechanical testing of the “Rhombic Dodecahedron” lattice structure demonstrated that the samples passed all tests without failures (Figs. 5 and 6). The hysteresis curve showed minimal energy dissipation, indicating that the sample withstood the load without undergoing mechanical changes.

SS curve and loading–unloading path for the cubic sample with “rhombic dodecahedron” structure. The samples withstand stress of 14MPa (a), and the cycled loading-unloading of load of 5100 N

The samples were then loaded at 5100 N and held for 1 h. After this loading cycle, the sample was dynamically cycled again, but the load was held relatively long.

As shown in Fig. 6, this multiple-loaded sample undergoes stress relaxation as the second cycle begins. A stress of about 0.64 MPa was estimated to be released in 200 s. This change can be seen in the hysteresis curve where the unloading curve deviates from the load path.

These changes indicate that there have been specific changes to the sample. Further loading destroyed the sample (Fig. 6).

Long-term loading – unloading path of “rhombic dodecahedron” cubic sample and hysteresis curve. Long-term loading – unloading path shows stress relaxation (a), which is evident in hysteresis of loading–unloading curve (b)

The modification of the lattice structure improved the mechanical performance, but the sample has not yet been qualified for use as an implant due to the risk of failure. The extreme load used in testing (5100 N) accelerated the failure and helped monitoring its mechanisms within a measurable time frame.

Further optimization was done by increasing the nodal diameter. Figure 7 shows the “rhombic dodecahedron” lattice structures with strut sizes of 2.5 mm. Both structures exhibit well-formed struts and none visible defects, unlike the thin node cases.

Actual 3D printed lattice structure; (a) the view, and (b) its microstructure

Figure 8a shows the SS curve of the “rhombic dodecahedron” sample loaded up to 5100 N. After loading, samples were held for 40 min and then dynamically loaded (Fig. 8b). Mechanical testing of the samples with increased strut sizes showed that they passed all tests without failures. This indicates that “rhombic dodecahedron” structures with strut sizes greater than 1.9 mm can be considered for implant design.

SS curve of the “rhombic dodecahedron”. (a) energy dissipation and (b) dynamic loading curves

To explore the shear loading response, a 45° shear stress was applied to the samples. According to the results, the “rhombic dodecahedron” withstood all loads and no damage was found. This led to the conclusion that the optimal structure of the talus might be the “rhombic dodecahedron”.

The patient was a 41-year-old male surgeon who sustained a traumatic fracture dislocation of his left talus in 2016 in a car accident. He is a smoker and overweight with no other known medical comorbidities.

His initial treatment consisted of open reduction and internal fixation (ORIF) with screw fixation within 24 h, which failed to heal. This led to the removal of the screws a year post-surgery, and the nailing the ankle from the calcaneus up which ended with an established AVN of the talus.

In 2018, an attempt was made to fuse the ankle joint by a calcaneal-tibial locked nailing. In 2019, with no fusion visible on X-rays and with pain, he had drilling of the talus and tibia plus distal dynamization of the rod by removing the calcaneus screws.

This did not lead to fusion. There was increasing and disabling pain, a stiff ankle and subtalar joint with deformity in the midfoot. Advice was given to obtain a Syme amputation. He refused that option and sought advice on a salvage procedure.

His surgeon consulted us for a foot-sparing solution to rid the patient of pain and allow ambulation without crutches. The plan was not to gain ankle motion but to ambulate without pain retaining his foot. The solution of a pan-talar fusion with triple arthrodesis was proposed and accepted by the patient. The patient was counseled on smoking cessation before his proposed surgical treatment, and for at least a year post-op to allow for the prosthetic 3-D printed implant to incorporate with the adjacent bones.

The staged approach was to first remove the locked nail that failed to fuse the ankle joint. This would be followed by implanting an antibiotic cement spacer, taking gap measurements of this stiff multiply operated ankle/foot.

The “Neo Talus” implant had to achieve not only ankle fusion, but also solid fixation with the tibia without subluxation or extrusion of the implant, fuse the subtalar joint and make sure there would not be a talo-navicular pain or subluxation with time. The design of the “neo talus” needed to have plates to the distal tibia – anteriorly and medially – and areas in its design to accommodate 2 screws fixation to each the calcaneus and the navicular. Additionally, a separate lateral plate would span from the tibia after resecting the distal fibula, cross-lateral to the “neo talus” and get affixed to the calcaneus.

In Fig. 9, AM manufactured customized implant was implanted into the excised talus space after the removal of the antibiotic cement spacer. The implant was filled with morselized bone graft obtained from the resected distal fibula (Fig. 9). The wounds were closed, and the leg and foot were immobilized in cast for a total of 6 weeks.

At 6 weeks post-op, he was transferred to a plastic molded ankle-foot orthosis (AFO) continuing non-weight bearing for 3 months. Partial weight-bearing was started, and by 5 months he was fully weight-bearing and walking with the AFO on it and no pain.

Figure 10 shows CT images of the patient’s ankle 14 months after surgery. It shows the designed implant perfectly fits the bony defect area. He remained painless, walking without canes or crutches at his last follow-up at 2 years post-op.

We have good early results and a happy patient who is back to working as a surgeon using both hands, standing, and walking without the need for crutches.

X-rays after insertion of implant intra-operatively; The implant is in a correct place, (a) front view, (b) side view

CT scans performed after 14 months: The implant is well aligned and in a good position (a) front view, (b) side view

We are cognizant that it is still too early to apply this approach to every case of AVN of the talus, and that there may be other ways to treat them. But in case of failures or more complicated deformed ankles and hindfeet, our solution seems to have addressed this patient’s need, returning him early to gainful employment in his profession as a surgeon.