Decreased femoral fracture load after cephalomedullary nail removal

Abstract

Aims

Spontaneous neck fractures are feared complications of cephalomedullary nail removal after successful healing of per- and subtrochanteric fractures. To date, the initial postoperative stability as well as the correct weightbearing regimen remain unclear. The aim of this biomechanical ex vivo study was to evaluate the initial postoperative failure load after hardware removal of specimens, which received cephalomedullary nails during their lifetime.

Methods

A total of 20 specimens of voluntary body donors were included in this study. Group 1 (n = 10) consisted of specimens that received cephalomedullary nails during their lifetime due to per- or subtrochanteric fractures. Each individual was matched for age, sex, femur size, and neck-shaft angle (Group 2 = control, n = 10). Biomechanical testing was performed in a single-leg stance setting, and volumetric bone mineral density (vBMD) was measured proximally at the femoral neck and distally at the epicondyles.

Results

Groups 1 and 2 differed significantly in terms of failure loads (p = 0.002), fracture types, and ratios of proximal and distal vBMD (p = 0.035). Femora after nail removal were significantly weaker (1,835.0 N vs 4,523.0 N) and showed lower ratios of proximal to distal vBMD (0.74 vs 1.18), which indicated altered stress distributions at the femoral neck in presence of femoral neck screws. They were further characterized by predominantly subcapital buckle-type fractures, while the control Group 2 showed predominantly transcervical fractures.

Conclusion

Altered stress distribution in presence of femoral neck screws leads to changes in biomechanical properties of the proximal femur, resulting in potentially unstable situations after nail removal in clinical settings. Elective removal of cephalomedullary nails should be undertaken with caution in view of the potentially increased fracture risk.

Cite this article: Bone Joint Res 2025;14(5):365–372.

Article focus

• Elective hardware removal of cephalomedullary femur nails is sometimes necessary, but may be accompanied by spontaneous and occult re-fractures of the femoral neck in the first weeks after surgery.

Key messages

• Elective removal of cephalomedullary nails cannot be recommended due to an increased risk of subcapital neck fractures, which might ultimately result in hip arthroplasty surgery.

• Implantation of cephalomedullary nails leads to structurally altered osseous situations, which have to be kept in mind when removing these devices.

Strengths and limitations

• This is the first time that femora of patients who received cephalomedullary nails due to a per- or subtrochanteric fracture during their lifetime were included in a postmortem biomechanical testing setup.

• Substantial differences in fracture loads, compared to studies in which nails were simply implanted postmortem in healthy bones, lead to the conclusion that these simple studies on healthy specimens might not show the whole story and need to be reviewed with scepticism.

Introduction

Cephalomedullary nails are used for internal fixation of proximal femoral fractures, as they allow for early postoperative mobilization in combination with full weightbearing.¹ The overall complication rate is low, but certain conditions such as pain, metal allergy, or patient wish may require hardware removal.^2,3 However, even after successful fracture healing, high rates of spontaneous refracture or the occurrence of new fractures at the subcapital femoral neck region after nail removal have been documented in various case reports,^4-7 and median fracture rates of up to 14.5% have been assumed.⁸ Correct management of these patients is challenging due to the obscure clinical presentation, with non-specific hip pain and no history of recent trauma.⁸ Treatment varies depending on the patient’s age, and includes total/hemi hip arthroplasty^4,7 or screw fixation.⁷ Since hip fractures are associated with increased morbidity and mortality,^9,10 various postoperative mobilization regimens were implemented to minimize the risk of potential fractures after elective hardware removal.¹¹ However, no consensus has been found to date about the correct approach.

Different systemic and implant-associated risk factors for re-fractures have been previously described. Systemic changes in bone mineral density (BMD) at the femoral neck, as well as different radiological parameters such as diameters of the neck screw or distances to the inferior and superior cortex, were associated with either increased or decreased fracture risks.^4,8,12 Based on Wolff’s Law,¹³ altered force transmission from the femoral head through the neck-screw into the femoral meta- and diaphysis could theoretically lead to cancellous bone loss at the femoral neck with increased fracture risks.¹⁴

Since previous anatomical and biomechanical studies used artificial setups of intact body donor specimens in which cephalomedullary nails were implanted postmortem, bone remodelling at the femoral neck has not yet been evaluated.¹⁵ Therefore, the aim of this ex vivo study was to evaluate the initial postoperative failure load of femora after hardware removal using specimens from body donors, who received a cephalomedullary nail during their lifetime due to a per- or subtrochanteric fracture. Our hypothesis was that femora after nail removal were statistically significant weaker compared to intact individuals. We further hypothesized that decreased BMD at the subcapital femoral neck region is the main factor responsible for the unique fracture pattern in these individuals.

Methods

Specimen acquisition

Overall, 20 body donor specimens were included in this ex vivo study (demographic data shown in Supplementary Table i). Group 1 (n = 10) consisted of femora after implantation of cephalomedullary nails during their lifetime with either a Stryker Gamma3 Hip Fracture Nailing System (Gamma3 Nail, n = 5; Stryker, USA) or a Synthes Proximal Femur Nail Antirotation (PFNA, n = 5; DePuy Synthes, USA) (Figure 1), and Group 2 (n = 10) of individually matched intact femora of other specimens based on sex, age, femoral length, and neck-shaft angle (Table I). All donors originated from the Vienna Medical Bio-/Implantbank of the Center for Anatomy and Cell Biology of the Medical University of Vienna, and provided informed written consent prior to their death to have their bodies used in medical education and research. The study was approved by the institutional review board of the Medical University of Vienna (2417/2020). Each individual was treated during their lifetime with a cephalomedullary nail due to a per- or subtrochanteric femoral fracture, and therefore providing a realistic representation of bone remodelling during fracture healing and beyond. All included specimens had the same neck-screw diameter of 10.5 mm. A visualization of the applied study protocol can be found in Figure 2. Inclusion criteria were available specimens with healed proximal femoral fractures with a minimum postoperative period of six months. The mean age of the implant carrier at the time of death was 87.9 years (SD 7.9), and the mean time between hardware implantation and death was 35.0 months (SD 22.7; 9.4 to 85.6).

Fig. 1.

Overview of included specimens. Upper row: DePuy Synthes Proximal Femur Nail Antirotation (PFNA) (USA). Lower row: Stryker Gamma3 (USA). L = long. Specimen 2 showed cement present within the femoral head surrounding the PFNA blade, and specimen 5 was excluded from biomechanical testing due to a nonunion at the pertrochanteric region. m, months.

Table I.

Demographic data (sex, age, femoral length, neck-shaft angle, and implant survival) for each group.

Group	Sex (F:M), n	Mean age, yrs (SD)	Mean femoral length, cm (SD)	Mean neck-shaft angle, ° (SD)	Mean implant survival, mths (SD)
Group 1 (n = 10)	9:1	87.9 (7.9)	42.8 (2.5)	123.4 (1.6)	35.0 (22.7)
Group 2 (n = 10)	9:1	87.8 (10.1)	42.7 (1.6)	124.8 (4.8)	Control

Fig. 2.

Workflow of the study protocol. BMD, bone mineral density; PFNA, Proximal Femur Nail Antirotation; vBMD, volumetric bone mineral density.

Specimen preparation

Each femur was explanted, macroscopically evaluated for deformities, and stored in vacuum-sealed bags at -20°C. First, each femur underwent x-rays and CT in a water bath to simulate surrounding soft-tissue, as previously described.¹⁶ One specimen of Group 1 was excluded due to a nonunion at the pertrochanteric region. In Group 1, cephalomedullary nails were then carefully removed according to the product portfolio of Stryker or DePuy Synthes. Osteotomies were performed in the middle of the femoral shaft, measured from the femoral head to the medial epicondyle. An additional 5 cm was kept distally for embedding in polyurethane (SG 141/PUR 145), and femora were then aligned 16° laterally in coronal and 9° posterior in sagittal plane to simulate physiological single-leg stance.¹⁷ To ensure correct alignment and to rule out iatrogenic fractures after nail removal, a second round of imaging with CT as well as radiographs was performed.

Bone mineral density and radiological analysis

Volumetric bone mineral density (vBMD) was measured in both groups at the epicondyles and at the femoral neck using custom Python scripts and Medtool 4.5 (Dr. Pahr Ingenieurs e.U., Austria). First, the CT scans after nail extraction were calibrated using a calibration phantom (QRM, Germany). Masks covering the entire bone volume were automatically generated using Medtool, but manually modified to erase bone fragments. Three landmarks were defined to allow a consistent definition of the regions of interest: the centre of the head and the centre of the smallest neck cross-section were automatically detected using a previously described method,¹⁸ and the centre of the epicondyles was found by manually fitting a cylinder to the distal articular curvature. To measure proximal and distal vBMD, the subcapital neck region and distal epicondyles were defined as illustrated in Figure 2, Step 3.

Additionally, standard anteroposterior radiographs with a 25 mm calibration ball of all specimens in Group 1 were taken. The following parameters were measured as they were associated with higher fracture risks:⁴ neck width in mm, relative implant width (neck width/screw diameter), and the distance from implant to the inferior cortex in mm.

Biomechanical testing

Since fractures after implant removal usually occur spontaneously and not because of a fall onto the hip,⁸ biomechanical testing was performed in a single leg stance setting without muscle forces as shown previously.¹⁹ Used device was a 30 kN electromechanical Zwick/Roell Z030 (Zwick/Roell, Germany) axial testing machine. Before testing, a metal cup was positioned onto the femoral head to allow a more uniform load distribution. A 25 kN load cell with 6° of freedom (Hottinger Baldwin Messtechnik (HBM) GmbH, Germany) was mounted on the crosshead. Hardened iron disks were used with ring ball-bearings to apply a purely axial load.²⁰ The crosshead was manually moved until a preload of 10 N was recorded. Ten pre-cycles of 100 N were applied, followed by axial loading with increments of 10 mm/min until failure was reached. Failure load was then defined as the maximum force during the testing procedure. Testing was further recorded with photographs and filming. After the biomechanical testing, each femur underwent a third round of imaging with CT scans in air and radiographs to evaluate fracture patterns (Figure 2, Step 5; for individual coronal images of each fracture pattern, see Supplementary Figure a).

Statistical analysis

The statistical analyses were performed with GraphPad Prism 9.1.0 for Windows (GraphPad Software, USA) and an Excel spreadsheet (Excel 365; Microsoft, USA). Descriptive statistics (mean, SD, range) were calculated for all metric variables (failure load, BMD, age, femoral length, neck-shaft angle, implant survival). Differences between Groups 1 and 2 were evaluated using an independent-samples two-sided t-test, and correlations between these variables were calculated using the Pearson correlation coefficient (r). Correlations were defined as follows: ≥ 0.9 excellent; ≥ 0.75 to 0.89 good; ≥ 0.5 to 0.74 moderate; and < 0.49 poor. A p-value < 0.05 was considered statistically significant.

Results

A total of 19 femora were included in the biomechanical testing. Except for one specimen, all included femora (n = 19) fractured either at the femoral neck (subcapital buckle type and transcervical), pertrochanteric region, or femoral shaft. A subcapital buckle-type fracture was defined as an instable minimally displaced complete subcapital fracture with a valgus tilt of the femoral head and a medial impaction zone.

Failure load

Mean failure load for femora after nail removal for Group 1 was 1,835.0 N (SD 669.5), and for Group 2 was 4,523.0 N (SD 2,161.0) (p = 0.002, independent-samples two-sided t-test). (Figure 3)

Fig. 3.

Force-displacement curves of all tested specimens. Group 1 (femora after nail removal) = black, Group 2 (intact control group) = grey. Mean failure load was 1,835.0 N (SD 669.5) for Group 1 and 4,523.0 N (SD 2,161.0) for Group 2. Differences were statistically significant (p = 0.002, independent-samples two-sided t-test).

Fracture type

In Group 1 (n = 9), there were six subcapital buckle type fractures, two transcervical fractures, and one femoral shaft fracture. For Group 2 (n = 10), seven transcervical fractures, one subcapital buckle-type fracture, one pertrochanteric fracture, and one case with no fracture were recorded. An overview of fracture types can be found in Figure 4.

Fig. 4.

Left: Diagram of fracture types at the proximal femur – subcapital fracture (dotted line), transcervical fracture (dashed line), and pertrochanteric fracture (xxx-line). The subcapital fracture was classified as a buckle-type fracture, i.e. an undisplaced complete subcapital fracture with a varus tilt of the femoral head and a medial impaction zone. Right: 3D volume renderings (upper row) and coronal CT reconstructions (lower row) of subcapital (left), transcervical (middle), and pertrochanteric (right) fractures found in this study.

Bone mineral density analysis

In Group 1, eight specimens and in Group 2 ten specimens were included in the vBMD analysis. One specimen of Group 1 had to be excluded due to cement in the femoral neck and head region. vBMD evaluation revealed a mean proximal vBMD of 97.82 mg/cm³ (SD 54.25) and mean distal vBMD of 129.37 mg/cm³ (SD 45.36) for Group 1. Group 2 showed significantly higher mean vBMD values at the proximal (199.41 mg/cm³ (SD 77.77)) and distal region (178.58 mg/cm³ (SD 71.72)). The mean ratio of proximal to distal vBMD was 0.74 for Group 1 and 1.18 for Group 2. This difference was statistically significant (p = 0.035, independent-samples two-sided t-test). For Group 2, there were good-to-moderate correlations between failure load and proximal/distal vBMD (p = 0.009 and p = 0.034, respectively), and for Group 1 there were moderate-to-weak correlations (p = 0.074 and p = 0.269, respectively; both Pearson correlation coefficient). Detailed results can be found in Table II.

Table II.

Results for volumetric bone mineral density (vBMD) for Group 1 (nail removal) and Group 2 (intact control), and correlations with failure load. The difference between Groups 1 and 2 for the vBMD ratio proximal/distal was significant (p = 0.035, independent-samples two-sided t-test).

Group	Mean vBMD proximal, mg/cm3 (SD)	Mean vBMD distal, mg/cm3 (SD)	vBMD ratio proximal/distal	Correlation failure load – vBMD proximal/distal (r, Pearson)
Group 1 (n = 8*)	97.82 (54.25)	129.37 (45.36)	0.74	0.66/0.44
Group 2 (n = 10)	199.41 (77.77)	178.58 (71.72)	1.18	0.76/0.66

^*One specimen was excluded due to a nonunion and one specimen due to cement in the femoral neck and head region, which hindered vBMD analysis (Figure 1).

vBMD, volumetric bone mineral density.

Evaluation of radiological parameters

The mean value for neck diameters was 30.6 mm (SD 3.0), and the mean distance from the implant to the inferior cortex was 8.2 mm (SD 1.8). No significant correlations to failure load were found for the neck width (r = 0.1), the relative implant width (r = 0.0), or the distance from the implant to the inferior cortex (r = 0.1).

Discussion

In this study, we investigated the initial femoral failure load after hardware removal of body donor specimens, which received cephalomedullary nails during their lifetime. They were compared to intact femora of other individuals without previous surgery, and differences in fracture patterns were investigated. Our main finding is that failure load after nail removal was statistically significant lower for femora after nail removal, leading to predominantly subcapital femoral neck fractures.

Nail removal after healed per- or subtrochanteric fractures might be rare, but is sometimes necessary due to infection, implant breakage, or non-specific hip pain. Multiple case reports described femoral neck fractures after elective removal of cephalomedullary nails.^4,14,21,22 Barquet et al⁸ conducted a systematic review to examine potential predictive factors for spontaneous fractures after implant removal. They concluded that pre-existing systemic osteoporosis as well as local osteoporosis resulting from altered force transmission may contribute to high fracture rates, estimated at 14.9% on average (median 14.5%) in a subset of four studies. Similar to our study, most fractures were found at the subcapital (95.55%) femoral neck region.

So far, the influence of implant time on the biomechanical performance of bone after implant removal has not been addressed. Yang et al¹⁵ compared the biomechanical stability of intact femora with femora that had a postmortem implanted PFNA-II removed. They found statistically significant lower failure loads in the nail group (4,085.6 N) compared to the intact control group (6,227.9 N). Solely trochanteric fracture patterns were found in the PFNA group. These findings stand in huge contrast to the results of our ex vivo study. While differences between failure loads in both studies for intact femora (4,523.0 N in this study vs 6,227.9 N in Yang et al¹⁵) can be attributed to their younger patient cohort, there are substantial differences for femora after nail removal (1,835.0 N vs 4,085.6 N). These differences can be explained by the different specimen selection; as already mentioned in the limitation section by Yang et al¹⁵ themselves, bony adaptions of fracture healing and altered force transmissions over time could not be taken into account, because they only included femora in which nails were implanted postmortem and not patients who suffered fractures during their lifetime.

Our main finding of statistically significant lower failure loads and altered fracture pattern of femora after nail removal can be primarily explained by the lower vBMD at the femoral neck region. While overall vBMD was also lower, the ratio between proximal and distal vBMD differed significantly between both investigated groups (0.74 vs 1.18, p = 0.03). Since the chosen cohort of femora with healed per- or subtrochanteric fractures already had classic osteoporotic fractures during their lifetime, lower overall vBMD was expected.²³ However, based on our findings, we believe that altered force transmission due to the femoral neck screw of cephalomedullary implants after successful fracture healing eventually leads to even lower vBMD specifically at the femoral neck. These conclusions can be supported by a characteristic type of subcapital buckle-type fracture seen in our study and other clinical case reports,⁸ which also differ from impacted femoral neck fractures that can be treated with simple femoral neck screws.²⁴

Our correlations of proximal vBMDs at the femoral subcapital neck region with failure loads were similar to the recent findings of Steiner et al.²⁵ Lower correlations for proximal vBMD and failure loads for femora after nail removal compared to intact bones without surgery emphasize the fact that bone remodelling after healed fractures in the presence of implants is complex. While the simple loss of bone after nail removal already leads to lower stability,¹⁵ the main factor responsible could be the statistically significant lower vBMD, most likely due to altered stress distribution in the presence of femoral neck screws. Contrary to some reports, no correlations for radiological parameters such as neck width, relative implant size, or distance to the inferior neck cortex could be observed in our patient cohort.^4,12

Clinical impact

Based on our findings, elective removal of cephalomedullary nails with femoral neck screws in elderly patients cannot be recommended, and should be reserved for special cases with clear indication. If absolutely necessary, we recommend a strict postoperative partial weightbearing regimen over a prolonged period, emphasized by the low mean failure load of 1,835 N found in our study. We also agree with the study by Driessen and Goessens¹¹ that, if possible, arthroplasty of the neck screw should be performed rather than simple removal. Whether divergent approaches would be suitable for younger patients cannot be clearly answered, as body donor specimens with pertrochanteric fractures during their lifetime under the age of 65 years, who donate their body to science, are rare and thus practically impossible to include in a sufficient number. As seen in our study, vBMD of the subcapital proximal neck region could be a potential – albeit weak – indicator for fracture risk. Since clinical studies in which different postoperative weightbearing regimens are compared might not be suitable, due to the drastically increased fracture risk as shown in our results, future studies of anatomical specimens with detailed CT scans might be necessary to bring more clarity and better predictors. Ideally, preoperative CT scans can be used to evaluate the postoperative fracture risk preoperatively.

Limitations

Several limitations must be considered for anatomical studies. First of all, due to the uniqueness of specimens with prior hip surgery, there was a low number of included specimens; however, this is in line with other anatomical studies.^16,26 Second, due to the study design of body donor specimens which received nails during their lifetime, only older patients could be investigated as previously outlined. However, by using ex vivo specimens, we were able to quantify the realistic failure load after successful fracture healing for proximal femoral fractures for the first time in the literature. Furthermore, we were not able to include the contralateral side of the same body donor as uninjured side control, as most donors had additional implants on the contralateral side. We therefore opted for a control group matched for sex, age, femoral length, and neck-shaft angle. Intra-individual differences of both femora could therefore not be evaluated, and comparisons in the present study were only made between patients who suffered a fragility fracture and relatively healthy specimens. A further experiment of postmortem retrieved patients with pre-existing fragility fractures and their uninjured contralateral side would of course further substantiate the results of the present study. Another limitation is the fact that a simple single-leg stance setup without consideration of muscle strength was used. Since this is still the standard approach for fracture analysis, muscle inclusion for head-neck fractures is not necessary,¹⁹ and in reality fractures after nail removal occur during single-leg stance and not due to a fall; we therefore believe that impactful conclusions can be drawn for clinical practice.

To the best of our knowledge, this is the first study to investigate femoral failure load after nail removal of body donor specimens, which received cephalomedullary nails during their lifetime. Femora after cephalomedullary nail removal were statistically significantly weaker compared to intact femora without previous surgery, and showed different fracture patterns. Altered stress distribution in the presence of femoral neck screws led to statistically significantly decreased vBMD at the subcapital neck region, resulting in predominantly subcapital buckle-type fractures, which could lead to unstable situations after nail removal in clinical settings.

Data Availability

The data that support the findings for this study are available to other researchers from the corresponding author upon reasonable request.