Initial stability of a percutaneous osseointegrated endoprosthesis with proximal interlocking screws for transhumeral amputees

Abstract

Background:

Percutaneous osseointegrated devices for skeletal fixation of prosthetic limbs have the potential to improve clinical outcomes in the transhumeral amputee population. Initial endoprosthesis stability is paramount for long-term osseointegration and safe clinical introduction of this technology. We evaluated an endoprosthetic design featuring a distally porous coated titanium stem with proximal slots for placement of bicortical interlocking screws.

Methods:

Yield load, ultimate failure load, and construct stiffness were measured in 18 pairs of fresh-frozen and thawed cadaver humeri, at distal and proximal amputation levels, without and with screws, under axial pull-out, torsion, and bending loads. Paired statistical comparisons were performed without screws at the two resection levels, and at distal and proximal levels with and without screws.

Findings:

Without screws, the location of the amputation influenced the stability only in torsional yield (p = 0.032) and torsional ultimate failure (p = 0.033). Proximally, the torsional yield and the torsional ultimate failure were 44% and 47% of that distally. Screws improved stability. In axial pull-out, screws increased the distal ultimate failure 3.2 times (p = 0.003). In torsion, screws increased the yield at the proximal level 1.9 times (p = 0.035), distal ultimate failure load 3.3 times (p = 0.016) and proximal ultimate failure 6.4 times (p = 0.013). In bending, screws increased ultimate failure at the proximal level 1.6 times (p = 0.026).

Interpretation:

Proximal slots and bicortical interlocking screws may find application in percutaneous osseointegrated devices for patients with amputations, especially in the less stable proximal bone of a short residual limb.

1. Introduction

Percutaneous osseointegrated (OI) devices for direct skeletal fixation of prosthetic limbs have been the subject of clinical trials in both upper and lower extremity amputees since 1990, with several endoprosthesis systems currently under investigation (Al Muderis et al., 2016; Aschoff and Juhnke, 2012; Brånemark et al., 2005; Hagberg and Branemark, 2009; Kang et al., 2010; Tsikandylakis et al., 2014). Percutaneous OI devices have demonstrated numerous advantages over socket suspension prosthetics, namely increased prosthetic use while obviating socket related complications such as discomfort, lack of heat dissipation, and inconvenience. Additionally, percutaneous OI devices have been shown to improve range of motion, performance of daily activities, and user satisfaction, while simultaneously offering improved sensory feedback through osseoperception (Branemark et al., 2014; Hagberg et al., 2008a; Hagberg et al., 2008b; Hagberg et al., 2014; Hagberg and Branemark, 2009). At the transhumeral level, percutaneous OI devices have been utilized as a conduit for enhanced neural control, and may improve efficiency in terminal device positioning (Ortiz-Catalan et al., 2014). Complications, including ascending infection at the percutaneous site, endoprosthesis loosening, and periprosthetic fracture have limited the clinical introduction of percutaneous OI devices with only one device to-date gaining FDA approval (HDE# H080004) for humanitarian use at the transfemoral level. While mitigation and clinical management of infections in percutaneous OI patients remains an active area of research, only one study has evaluated primary stability of the endoprosthetic stem, which is a prerequisite for safe clinical adoption of this technology (Welke et al., 2013).

Current endoprostheses under investigation in the transhumeral population use diverse means of skeletal fixation, including an intramedullary screw (Tsikandylakis et al., 2014), a hydroxyapatite coated press-fit stem with radial splines (Kang et al., 2010), and a plasma spray coated endoprosthesis that utilizes an intramedullary anchor plug and pre-stressed washers to apply high compressive load to the osteotomy (McGough et al., 2017). While some of these devices trace their lineage to either dental or oncologic reconstruction applications, we propose an alternative that draws inspiration from the titanium porous coatings used in cementless total joint arthroplasty where distal ingrowth fixation combined with a proximal stabilizing stem are associated with excellent long-term survivorship (Khanuja et al., 2011). The principles of this endoprosthetic design were previously validated in a load bearing ovine model (Jeyapalina et al., 2012; Jeyapalina et al., 2014b; Jeyapalina et al., 2014a; Shelton et al., 2011a) and the approach has been expanded to clinical trials (Clinical Trials # NCT02720159) in transfemoral amputees.

Both the ovine and transfemoral stem utilized proximal splines for endosteal engagement and enhanced torsional stability. In the funnel-shaped diaphysis of the humerus (Drew et al., 2019), engagement of such features is challenging due to the diverging endosteum of the humeral diaphysis, which requires an alternative means of proximal stabilization. Also, the primary compressive loading regime of the lower extremity (Frossard et al., 2013; Lee et al., 2008) is replaced in the upper extremity with combined axial distractive, torsional, and bending loads (Drew et al., 2017; Stenlund et al., 2019), requiring additional features for initial stability not afforded by the lower extremity designs.

In this study, we evaluated the initial stability of a press fit, porous coated, end loading transhumeral percutaneous OI device with proximal slots for placement of bicortical interlocking screws. Specifically, we evaluated the difference between proximal and distal amputation levels to look at the effects of the changing cortical thickness and medullary shape (Drew et al., 2019) on initial stability. Proximal amputations will likely experience lower press fit forces due to thinner cortex and steeper flare (Drew et al., 2019). We also examined whether the use of proximal interlocking screws, similar to those used for intramedullary nail fixation, improved primary stability of the endoprosthetic stem. Yield and ultimate failure loads, as well as construct stiffness, were evaluated under axial pull-out, torsion, and bending load configurations, at both a distal and a proximal anatomical location, with and without the use of screw fixation.

There were two goals of this study. First, the initial stability of a press fit, porous coated, and end loading transhumeral percutaneous OI device was evaluated. We hypothesized that distal amputation levels would have higher yield and ultimate failure loads and stiffness than proximal levels. Second, the use of proximal interlocking screws to enhance the initial stability was evaluated. We hypothesized that proximal interlocking screws would increase yield and ultimate loads, as well as the construct stiffness, under each test condition at each anatomic location.

2. Methods

2.1. Design and manufacture

The evaluated percutaneous OI design features a 60 mm endoprosthetic stem comprised of a proximal cylindrical portion and a conical distal region with a titanium porous coating currently utilized in total hip, knee, and shoulder arthroplasty applications (P², DJO Surgical, Austin, TX) (Fig. 1). The proximal cylindrical region has two slotted features oriented 50° apart for placement of bicortical interlocking screws. Interlocking screws were placed in the proximal aspect of the slot, and as a result are exposed to mechanical loading along all axes except for axial compression. This configuration is an attempt to mitigate distal stress shielding during possible compressive end loading in-vivo. Five sizes (12–16 mm) were manufactured for this investigation. The distal end of the endoprosthesis was modified to include a square cross section and transverse hole for engagement with test fixtures.

Fig. 1.

Rendering of the proposed transhumeral percutaneous OI device featuring a distal press-fit, porous coated endoprosthetic stem with proximal interlocking screws for enhanced stability.

2.2. Implantation

Endoprosthetic stems were implanted in 18 pairs of fresh-frozen and thawed cadaver humeri. Each bone was first amputated distally at a location 70% of biomechanical length (BML) from the humeral head center. The bone was secured in a fixture for preparation of the medullary canal and distal osteotomy with surgical instrumentation. Preparation involved the use of straight reamers to establish the canal axis and ensure clearance of the proximal endoprosthetic stem. Tapered reamers with an integrated planer were then used to simultaneously prepare the endosteal bone and produce an orthogonal osteotomy plane for proper seating of the porous coated, end loading collar. An additional 5% BML was removed during preparation, resulting in a final distal osteotomy length of 65% BML. A broach corresponding to the selected endoprosthesis size was then used under manual power for final preparation of the canal. Broaching continued until trial placement of the press-fit stem left the endoprosthetic collar approximately 3 mm proud of the distal osteotomy. This distance was determined empirically to reduce the risk of intraoperative fracture while ensuring tight press-fit of the porous coated stem. Final impaction was carried out with a surgical mallet. Finally, screw placement was randomized between left and right humeri so that each pair had one side implanted with screws and the other solely relied on the press-fit of the endoprosthesis (no-screw configuration). For humeri assigned screws, two 5.0 mm diameter screws were placed in the proximal aspect of the endoprosthesis slots with the assistance of a targeting guide (Fig. 2).

Fig. 2.

3D printed prototype targeting guide for screw placement. The guide was designed to ensure bicortical screw placement in the proximal aspect of the slotted feature present in the cylindrical region of the stem.

Prior to any mechanical testing, the specimens were then re-amputated proximally at 35% BML (Fig. 3). The bone was prepared as described above, resulting in a final proximal osteotomy length of 30% BML, followed by implantation of the press-fit stem. Therefore, each humerus subsequently was implanted with an endoprostheses placed at 65% BML (distal level) and at 30% BML (proximal level), representing a long and a short residual limb, respectively. Anteroposterior (AP) and mediolateral (ML) plain radiographs were collected following each implantation to qualitatively assess fit (Fig. 3).

Fig. 3.

Intact pairs of fresh-frozen and thawed human humeri were initially amputated at 70% BML, then implanted with an endoprosthesis to 65% BML (center). Next, each humerus was re-amputated at 35% BML, then implanted with another endoprosthesis to 30% BML (left and right). By re-amputating the specimen proximal to the endoprosthesis, two separate specimens per humerus, one at the distal level and one at the proximal level, were created prior to mechanical loading, minimizing any potential influence of testing each specimen at two amputation levels. Images show implantation into paired humeri with and without proximal interlocking screws (left/right assigned randomly). Inset shows the desired proximal screw placement in the proximal aspect of the endoprosthesis slot to improve initial stability in pull-out, in torsion, and in bending.

2.3. Axial pull-out

Implanted humeri at both the distal and the proximal levels were positioned in a servo-hydraulic material test machine (Model 858 Mini Bionix II, MTS Systems, Eden Prairie, MN, USA) equipped with a 25 kN load cell (#622.2OH-05, MTS Systems) such that loading occurred along the long axis of the endoprosthetic stem (Fig. 4). The endoprosthesis was affixed to the actuator, and the humerus was lowered into a clamping fixture and secured with set screws around the proximal anatomy beyond the proximal tip of the endoprosthetic stem. A 10 N tensile preload was applied to the construct, which was then distracted under displacement control at a rate of 5 mm/min until failure (Jeyapalina et al., 2014b). Force and displacement data were acquired at 1 kHz.

Fig. 4.

Schematic of axial pull-out, torsion and bending test conditions. Pull-out (left) and torsion (center) were tested along the long axis of the endoprosthesis. Bending (right) was applied to the cortical surface in-line with the bicipital groove by an aluminum roller. Load was applied 80 mm from the osteotomy.

2.4. Torsion

Implanted humeri were positioned in the material test machine (Model 858 Mini Bionix II, MTS Systems, Eden Prairie, MN, USA) equipped with a 250 Nm load cell (#622.2OH-05, MTS Systems) such that torque was applied about the long axis of the endoprosthetic stem (Fig. 4). Specimens were loaded into the machine as described previously for axial pull-out. A 0.5 Nm preload was applied to the construct, which was then loaded in torsion under displacement control at a rate of 1°/sec until failure.

2.5. Bending

Implanted humeri were secured perpendicular to the load axis of a servo hydraulic materials testing system (Model 8500, Instron Corp, Canton, MA, USA) equipped with a 5 kN load cell (#2518–103, Instron) by the distal square adapter on the endoprosthesis. Load was applied by an aluminum roller to the cortical surface of the bone at a distance 80 mm from the distal, porous coated, end loading collar (Fig. 4). Care was taken at the time of implantation to align the distal aspect of the stem such that bending loads were consistently applied in line with the bicipital groove of the humerus, resulting in a posteriorly directed moment. A 10 N bending preload was applied, which was then loaded under displacement control at a rate of 5 mm/min until failure.

2.6. Data analyses and statistics

In total, 72 tests were carried out (18 pairs × 2 sides × 2 amputation levels). Constructs were tested in axial pull-out, torsion, and bending (n = 24 per test configuration) (Fig. 4). Yield load was defined as the force or torque necessary to achieve 150 µm of endoprosthesis displacement, as this has been identified as a threshold above which sustained micromotion leads to mature fibrous encapsulation and endoprosthesis loosening (Ducheyne et al., 1977; Engh et al., 1992; Kienapfel et al., 1999; Pilliar et al., 1986). In torsion, this was measured by 150 µm radial arc calculated based on the nominal size of the selected endoprosthesis. In axial pull-out and bending, this was measured by a 150 µm displacement of the test machine crosshead. Ultimate failure load was calculated as the peak of the force or torque-displacement curves. Stiffness was calculated as the slope of the pre-failure linear region of the force or torque-displacement curve.

Statistical comparisons were carried out using 2-tailed, paired t-tests within specimens (with or without screws) at the two resection levels, and between specimen pairs with and without screws at a given resection level. p-Values < 0.05 were considered significant.

3. Results

Humeral specimens had a median age of 53 years (range: 19–74). Donors included 14 Caucasian males, 1 Hispanic male, 2 Caucasian females, and 1 African American female. Endoprosthetic size utilization skewed toward 12 mm stems (33 used). Utilization for the remaining endoprostheses was 20, 13, 4, and 2 for sizes 13–16 mm, respectively. Endoprosthesis utilization in the proximal diaphysis was (median) 1 mm larger than the size used for the distal diaphysis. Post-operative radiographic assessment by implanting surgeons showed acceptable apposition of the porous coated region with endosteal bone in all cases. Ten total intraoperative fractures occurred in 72 implantations. All fractures occurred during final broaching of the medullary canal or stem impaction. Paired specimens including at least one intraoperative fracture were excluded from statistical comparison if the fracture occurred in the no-screw endoprosthesis configuration. This was observed to significantly decrease the load supported by the bone-endoprosthetic interface due to loss of the press fit fixation.

3.1. Axial pull-out

Five of the ten intraoperative fractures were observed in this subset, resulting n = 5 and n = 4 paired comparisons at the distal and proximal levels, respectively. In the no-screw configuration, the yield load, ultimate failure load, and construct stiffness were not significantly different between the distal and the proximal locations (Table 1, p ≥ 0.117). Addition of interlocking screws did not significantly change the yield load between the distal or proximal levels (p ≥ 0.223). Ultimate failure load at the proximal level did not demonstrate a difference between the no-screw and screw configurations (p = 0.940), but a significant change in ultimate failure load was observed at the distal level with the use of screws, where failure occurred at 1594.8 N (SD 841.6) and 5120.3 N (560.7) for the no-screw and screw configurations, respectively (p = 0.003). No significant change in construct stiffness was observed with the use of interlocking screws at either the proximal or distal levels (p ≥ 0.697).

Table 1.

Mechanical properties during axial pull-out tests.

Axial pull-out	Stem without screws		Stem with screws
Mean (SD)	65% BML Distal	30% BML Proximal	65% BML Distal	30% BML Proximal
Force (N)
Yield	1364.8 (420.7)	784.2 (236.2)	1818.1 (393.5)a	805.1 (391.8)a
Ultimate	1594.8 (841.6)b	1325.1 (185.8)	5120.3 (560.7)a,b	1529.7 (702.3)a
Stiffness (N/mm)	11,222.1 (1323.3)	5797.7 (2272.7)	12,062.7 (2887.4)	6248.7 (3033.0)

Excluding humeral shaft fractures during stem placement (n = 3 of 24).

^aStatistically significant differences: distal vs. proximal.

^bStatistically significant differences: stem vs. stem + screws.

3.2. Torsion

Three of the ten intraoperative fractures were observed in this subset resulting in n = 6 and n = 5 paired comparisons at the distal and proximal levels, respectively. In no-screw configurations, there were significant differences between proximal and distal levels in yield (Table 2, p = 0.032) and ultimate failure (p = 0.033), but not in stiffness (p = 0.187). No improvement in yield load was observed distally when screws were added (p = 0.693). Yield load at the proximal level was enhanced 1.9× with the use of interlocking screws (p = 0.035). Ultimate failure load significantly increased with the addition of interlocking screws at both distal (p = 0.016) and proximal (p = 0.013) levels. No change in construct stiffness was observed with the use of interlocking screws at either the distal or proximal levels (p ≥ 0.339).

Table 2.

Mechanical properties during torsion tests.

Torsion test	Stem without screws		Stem with screws
Mean (SD)	65% BML Distal	30% BML Proximal	65% BML Distal	30% BML Proximal
Moment (Nm)
Yield	10.2 (3.8)a	4.5 (2.1)b,a	9.6 (5.3)	8.6 (2.8)b
Ultimate	12.7 (5.2)b,a	6.0 (2.6)b,a	42.2 (20.6)b	38.6 (14.1)b
Stiffness (Nm/deg)	9.0 (2.1)	6.3 (2.5)	8.8 (3.1)	8.8 (2.7)

Excluding humeral shaft fractures during stem placement (n = 3 of 24).

^aStatistically significant differences: proximal vs. distal.

^bStatistically significant differences: stem vs. stem + screws.

3.3. Bending

Two of the ten intraoperative fractures were observed in this subset resulting in n = 6 and n = 5 paired comparisons at the distal and proximal levels, respectively. In no-screw configurations, no significant differences were seen between distal and proximal levels in yield load, ultimate failure load, or stiffness (Table 3, p ≥ 0.131). No measurable difference was observed in yield load at the either the distal level (p = 0.203) or at the proximal level (p = 0.073) with the addition of screws. Likewise, no measurable difference was observed in the ultimate failure load at the distal level (p = 0.213). However, the addition of screws provided a significant change in ultimate failure load at the proximal level (p = 0.026). No change in construct stiffness was observed with the use of interlocking screws at the proximal and distal levels (p ≥ 0.097).

Table 3.

Mechanical properties during bending tests.

Bending test	Stem without screws		Stem with screws
Mean (SD)	65% BML Distal	30% BML Proximal	65% BML Distal	30% BML Proximal
Moment (Nm)
Yield	2.0 (0.6)	1.7 (0.5)	2.3 (0.2)	2.2 (0.5)
Ultimate	83.8 (37.5)	70.3 (37.6)a	119.4 (37.1)	111.8 (18.3)a
Stiffness (Nm/mm)	7.7 (1.9)	5.5 (1.9)	12.9 (0.4)	11.9 (5.5)

Excluding humeral shaft fractures during stem placement (n = 2 of 24).

^aStatistically significant differences: stem vs. stem + screws.

4. Discussion

The first goal of this investigation was to evaluate the initial stability of a press fit, porous coated, end loading transhumeral percutaneous OI device at different amputation levels. We hypothesized that amputations at the distal level would have higher yield and ultimate failure loads, and stiffness. Though the results of this study were in agreement with this trend, there was no consistent significant difference between yield load, ultimate failure load, or stiffness at the distal and proximal amputation levels in axial pull-out and bending. In torsion, the yield and ultimate failure loads distally were significantly higher than proximal by 2.2× and 2.1×, respectively. This is likely due to the increased cortical thickness and more gradual medullary flare at the distal level, increasing cortical contact at the porous coated region (Drew et al., 2019).

Axial pull-out in a load bearing ovine model of a similar porous coated endoprosthetic OI system showed time zero pull-out strength of 988 N (874). This strength increased to 13,485 N (1626) after 12 months in situ (Jeyapalina et al., 2014b). Results presented herein show at least a 1.5× increase in ultimate pull-out strength compared to time zero ovine data. We expect this to improve in patients over time with increased bone ingrowth similar to the ovine model (Jeyapalina et al., 2014b).

The second goal of this investigation was to examine if proximal interlocking screws enhanced initial endoprosthetic stability. The hypotheses tested were that proximal interlocking screws would increase yield, ultimate loads, and construct stiffness, under each test condition. Significant increases in yield and ultimate loads were observed with the use of proximal screws for some loading conditions when compared to press-fit fixation alone. Axial pull-out results for yield load were unaffected by proximal screw use, indicating the screws did not consistently engage the endoprosthesis in the slot within the 150 µm yield range. This hypothesis was not supported for stiffness across all tests, but was partly supported in axial pull-out and torsion. Stiffness was not significantly altered when comparing proximal and distal amputation levels, or use of screws in all tests. This result indicates that at 150 µm of displacement, the contribution of the screws to stiffness of the construct is not realized and stability at the micromotion scale is dominated by the press-fit engagement of the distal porous ingrowth surface.

When looking at the entire axial pullout data set collected throughout this study, combining the yield loads for the distal and the proximal levels, and the loads with and without screws, data ranged from 405.8 N to 2038.6 N. These values are greater than those measured in a cohort of transhumeral amputees fit with a percutaneous OI device (Stenlund et al., 2019). In their study, mean maximum longitudinal (axial) loads observed during activities of daily living ranged from 24.1–109.1 N (Stenlund et al., 2019). These measured loads were comparable to the range of transhumeral axial loads previously estimated by inverse dynamic analysis in non-amputee subjects, 37.5–138.7 N (Drew et al., 2017). While mean axial yield loads reported herein were over 5× greater than mean maximum loads in previous studies (Drew et al., 2017; Stenlund et al., 2019), it should be noted that a limited range of daily activities have been evaluated in the literature and none have examined traumatic or uncontrolled events.

Torsional yield load in the proximal humerus saw a nearly 2× improvement with use of proximal screws, indicating that screw placement stabilizes the press-fit contact of the endoprosthesis in the proximal humerus where the funnel shaped diaphysis impacts overall bone-endoprosthesis contact area. Torsional yield loads for all configurations with and without screws exceed those reported for elbow supination/pronation by Murray et al. during ADLs (0.05 Nm), but are lower than values for some ADLs associated with moderate demand activities (Drew et al., 2017; Murray and Johnson, 2004; Stenlund et al., 2019), which reached single subject maximums as high as 57.2 Nm (Drew et al., 2017). While peak predicted axial force of ADLs was much lower than yield loads for any configurations with and without screws, this did not hold true for torsional and bending moments (Drew et al., 2017). Here, the mean torsional yield and ultimate failure loads and mean bending yield values fell below predicted and measured ADL torsion and bending moments (Drew et al., 2017; Stenlund et al., 2019). This indicates that early prosthetic rehabilitation may be tolerated with this design, but activities should be restricted or protective countermeasures, such as overload protection, should be employed at the prosthetic connection. The range of ultimate torsional failure loads observed in this investigation when proximal screws were utilized (15.3–64.1 Nm) agrees with reported periprosthetic fracture loads at the shoulder (5.3–23.4 Nm) (Flury et al., 2011) and elbow (16.5–79.3 Nm) (van der Lugt et al., 2010), further motivating the need for overload protection against this failure modality.

Ultimate bending failure loads can be directly compared to the work of Welke et al. which evaluated the bending loads of a commercially available, cementless, endoprosthetic stem of similar diameter (Welke et al., 2013). Ultimate failure loads in that investigation were approximately one-third the mean value attained with screw use at both amputation levels, 36.7 Nm (11.0). Variation in cadaver specimens may be a significant contributor to this result, since the higher median age (73.5 years) reported in their investigation could have been associated with a decrease in bone mineral density and lower strength of the specimens.

The observed rate of intraoperative fracture (14%) may be attributed in part to the limited range of endoprosthesis sizes available for this investigation. Additionally, more fractures were seen in axial testing than torsion or bending which could be attributed to the fact that axial endoprostheses were placed first and there was a learning curve for the surgeons placing the system. The observed rate of fracture is lower than that reported for transhumeral use of the OPRA system (44%) although disuse osteoporosis in amputee patients likely contributes to this high rate of occurrence (Tsikandylakis et al., 2014). When compared to arthroplasty applications in the upper extremity, the fracture rate is similar to revision total shoulders where 16% of humeri experience intraoperative periprosthetic fracture (Wagner et al., 2015). It is also important to note that all endoprostheses were placed in thawed cadaver bone which may be less hydrated and exhibit decreased energy to fracture compared to in vivo bone.

This study has several limitations. First, is the use of cadaver bones that do not capture the deficient bone quality possible with actual residual limbs. For this reason, we expect the loads observed herein to be higher than those for long-term amputees, highlighting the importance of the use of screws to enhance initial stability of the stem. Second, since each cadaver bone was tested at both distal and proximal amputation levels, it is possible that the distal implantation caused damage that could affect the results of the proximal amputation level. However, this potential limitation was minimized by re-amputating the specimen proximal to the endoprosthesis prior to mechanical testing of the distal level, making essentially two separate specimens per humeri. Third, this study looks at initial fixation only, modeling initial post-operative fixation. No attempt was made to simulate bone ingrowth or bone remodeling effects. Future investigations of percutaneous OI initial stability should include screening of cadaveric tissue by either dual-energy x-ray absorptiometry (DEXA), or quantitative CT to select for specimens with bone density matched to the expected ranges of amputee populations. Future studies should also include the full range of available endoprosthesis sizing, whereas the present study utilized less than half the expected range. Undersized endoprostheses likely compromise initial fixation due to the reduced area of distal porous coating contact and resultant hoop stress during impaction, whereas oversized endoprostheses could result in excessive bone removal and consequent cortical thinning during preparation. Both scenarios are likely to decrease the forces supported at the bone-endoprosthesis interface and increase the importance of the screws in providing initial stability.

5. Conclusion

We evaluated a transhumeral percutaneous OI device that utilizes proximal interlocking screws for improved primary stability of the bone-endoprosthesis interface. No difference by amputation level was seen in axial pull-out and bending tests. Distal amputation levels were stronger in torsion. The use of screws resulted in a 3.2× increase in ultimate failure load under axial pull-out at the distal amputation level. Increased stability from screw use was also observed under torsional loading, with significant increases in ultimate failure load at both the distal and proximal levels. Importantly, torsional yield loads were also improved proximally where the pronounced metaphyseal flare of the endosteum makes proximal stabilization of the stem challenging. These results indicated that this design strategy is a viable approach to percutaneous OI at the transhumeral level and should be considered for clinical evaluation.