Comparison of reamed long and short intramedullary nail constructs in unstable intertrochanteric femur fractures: A biomechanical study.

OBJECTIVES: There is no definitive evidence to guide clinicians in their decision-making for implant choice regarding long or short intramedullary nails for unstable fracture patterns. Historically short nails were associated with higher rates of perisprothetic fractures which seem to have improved with newer designs. Long intramedullary nails have higher blood loss and time under anesthesia. The purpose of this study was to assess stability of long and short intramedullary nail constructs in unstable intertrochanteric fracture patterns to better elucidate if unstable intertrochanteric fractures are amenable to treatment with short intramedullary nails.
METHODS: This study utilized composite model femurs which were assigned to either a comminuted or reverse obliquity fracture pattern, then subsequently assigned to implantation with either a long or short intramedullary nail. All the samples were reamed to the level of the distal femur and instrumented with the appropriate nail. Axial and torsional stiffness as well as axial load to failure values were determined using a servohydraulic loading system.
RESULTS: Short nail constructs exhibited significantly greater axial stiffness in A1 fractures and torsional stiffness in A3 fractures when compared with long nails. There was no significant difference between axial load to failure between long nails and short nails. DISCUSSION: We found no significant difference in axial load to failure values between long and short intramedullary nail fixation in 2 unstable intertrochanteric fracture patterns in a composite femur model. Short nails exhibited greater stiffness in axial loads in the A1 pattern and torsional stiffness in the A3 pattern. This suggests short or long intramedullary nails could be appropriately employed for fixation of unstable intertrochanteric hip fracture patterns.

Objectives

Hip fractures are a large and growing health issue in the United States. In 2010, there was an incidence of 258,000 hip fractures in the United States, which totaled 17 to 20 billion dollars in associated healthcare costs.[ Intertrochanteric fractures account for roughly half of all hip fractures in the United States annually. Both long and short intramedullary nail (IMN) constructs are considered acceptable forms of fixation for intertrochanteric fractures. However, there is no definitive evidence to guide clinicians in their decision-making for implant choice in unstable fracture patterns.[ There is concern that short nails may not offer sufficient stability distal to the fracture site in these patterns, and they have been historically associated with increased incidence of subsequent periprosthetic fracture.[ Long intramedullary nails are thought to offer increased stability in unstable patterns.[ However, this added stability comes with a longer time under anesthesia[ and higher blood loss,[ both of which can increase morbidity in this fragile patient population. A series of studies have evaluated differences in intertrochanteric fractures managed with long and short intramedullary nails,[ including clinical outcomes as well as biomechanical properties in both cadaveric and synthetic femur models. However, many of these studies are underpowered and do not differentiate outcomes for unstable intertrochanteric fracture patterns or the different generations of short nails utilized in their study. As a result, clinicians lack sufficient evidence to guide their decision making for implant choice in patients with unstable intertrochanteric fracture patterns.[ In a recent survey regarding selection of long or short nail constructs for management of intertrochanteric fractures, 29% of surgeons indicated they use only long nail constructs citing concern for peri-implant fracture with use of short nails and biomechanical superiority of long nails as the 2 primary reasons for their rationale.[

The purpose of this study was to assess stability of long and short IMN constructs in 2 unstable intertrochanteric fracture patterns by assessing the axial load to failure and construct stiffness for each construct using a composite femur model. This information will better elucidate if unstable intertrochanteric fractures are amenable to treatment with short IMN or if they require the potential benefits and added stability of long IMN. We hypothesized that short IMN constructs would not be inferior to long IMN for fixation of unstable intertrochanteric fractures.

Methods

Nail implantation and fracture simulation

Our study utilized 4th generation Medium Femur composite sawbones models with a 13 mm canal diameter and a low density (10 pcf) cancellous bone model with a thin-walled cortical shell (Pacific Research Laboratories, Vashon Island, WA, SKU: 3403-106) to more closely mimic osteoporotic bone. Four treatment groups consisting of 8 constructs in each group were created. Each treatment group was assigned to either a short or long IMN group and designated to either comminuted (AO 31-A1.3) or reverse obliquity (AO 31-A3.1) fracture groups (Table 1).

Table 1

Overview of experimental groups

To create the fracture patterns, a custom cutting jig was made to standardize the osteotomy on each sample. The comminuted (A1) osteotomy extended from the middle of the greater trochanter to the tip of the lesser trochanter (LT) with an additional 3 cm portion of bone removed from the medial aspect which included the LT (Fig. 1A). The reverse obliquity osteotomy was made starting at the most proximal part of the LT and traveled distally and laterally, exiting at a point 1 cm below the most distal aspect of the LT on the lateral cortex (Fig. 1B). These fracture patterns were selected for testing because they represent 2 commonly seen unstable intertrochanteric fracture patterns and are currently commonly managed with long intramedullary nail constructs.[

Figure 1

(A) Demonstrates osteotomy pattern for A1 fracture models, extending from the middle of the greater trochanter and extending distally and medially to the tip of the lesser trochanter with an additional 3 cm portion of bone removed from the medial aspect which included the lesser trochanter. (B) Demonstrates the osteotomy cuts for A3 fracture models which begin at the most proximal part of the lesser trochanter and travel distally and laterally, exiting at a point on the lateral cortex 1 cm below the most distal aspect of the lesser trochanter.

With the aid of fluoroscopic guidance, the appropriate IMN (Natural Nail, Zimmer Biomet, Warsaw, IN) was inserted into the osteotomized femurs. To prepare the femurs for implantation, the proximal femur was reamed using a 15.5 mm opening reamer to the level of the LT. The shaft was then reamed to 13 mm. An 11.5 mm diameter nail of the selected length was then inserted followed by a 10.5 mm diameter lag screw. A set screw was placed and fully tightened in such a way that it created a fixed angle device, preventing any further compression. Finally, a single, static 5.0 mm distal interlock screw was placed in each model. Tip apex distance (TAD)[ was measured and recorded for each specimen following implantation to ensure appropriate lag screw positioning, as it has been shown that TAD (sum of the distance measured from the tip of the lag screw to the apex of the femoral head on AP and lateral fluoroscopic views) in excess of 2.5 cm is a risk factor for lag screw cut out.[

Prior to biomechanical testing, the distal aspect of each specimen was constrained in a polymethyl methacrylate split mold with vertical orientation so that both femoral condyles were touching the bottom of the mold.

Mechanical testing

Cyclic loading and axial load-to-failure tests were conducted on an Instron 8511 servohydraulic testing machine (ITW, Norwood, MA). For cyclic loading, load and displacement data was recorded at 25 Hz through a LabView Data Acquisition program (National Instruments, Austin, TX). The proximal end of the femur was not constrained during testing. For ultimate failure testing, the same system was used to measure at a rate of 100 Hz. Stiffness calculations and data analysis were conducted using R (The R Foundation for Statistical Computing, Vienna, Austria). We did not measure side loads or rotation/translation of the femoral head during testing.

This study was calculated to be adequately powered (power = 0.80, alpha 0.05) to detect differences in 371N in the ultimate axial load to failure testing with a sample size of 8.

Construct stiffness under axial compression was measured by placing the specimen in an upright orientation with distal aspect of the condyles parallel to the floor. Loads were applied ranging from 100 to 500 N of compression through the femoral head for 10 repetitions at 0.2 Hz (Fig. 2A). Stiffness was calculated as the average slope of the compressive segment of the load–displacement curve.

Figure 2

Mechanical loading setup for measurement of (A) axial stiffness and axial load to failure and (B) torsional stiffness.

Offset torsional stiffness was mimicked by placing each construct in a horizontal orientation with the distal end secured in a vice and a fulcrum placed 2 cm distal to the lesser trochanter to reduce 3-point bending force across the diaphysis. A load was then applied across the femoral head in a posterior direction causing internal rotation of the femoral diaphysis (Fig. 2B). Each construct was loaded from 50 to 150 N for 10 repetitions at 0.2 Hz.

Order of torsional and axial loading was randomized to account for the fact that the initial testing may affect the results of the subsequent testing. Each construct was then loaded to failure in axial compression at a rate of 5 mm/min with a preload of 100 N to assess mode of failure for each group. The maximum load applied prior to failure of the construct was considered the failure load. Rates and modes of failure for all constructs were observed and recorded.

Results

Tip apex distance

All TAD measurements were within Baumgaertner et al's[ recommended limit of less than 2.5 cm (range 0.18 cm to 2.41 cm). TAD was similar between groups with A3 fractures (P = 0.5504), but lower in the group fixed with long nails with A1 fracture pattern when compared with short nail A1 fracture (SNA1 15.9 mm, LNA1 10.9 mm), (P = 0.0474). There was no significant difference in TAD overall between A1 and A3 fracture groups (P = 0.6949).

Axial stiffness

Short nail constructs exhibited greater axial stiffness in A1 fractures (637 N/mm) compared with long nails (527.2 N/mm), (P = 0.0030). There was no significant difference in axial stiffness between short and long nail constructs for A3 fractures. Overall, A1 fractures exhibited lower axial stiffness (582.1 N/mm) when compared with A3 fractures (783.2 N/mm), (P = 0.0004) (Fig. 3).

Figure 3

Mean axial stiffness. Horizontal bars at the top indicate pairs of constructs with significantly different values (P < .05). Error bars indicate standard deviation.

Torsional stiffness

Overall, A1 fractures exhibited a higher torsional stiffness (79.8 N/mm) when compared with A3 fractures (67.8 N/mm), which was a significant difference (P = 0.0306). Additionally, short nails exhibited significantly higher torsional stiffness in A3 fractures when compared with long nails in A3 fractures (P = 0.0028). There was no significant difference between short nails and long nails in the A1 fracture pattern group (Fig. 4).

Figure 4

Mean torsional stiffness. Horizontal bars at the top indicate pairs of constructs with significantly different values (P < .05). Error bars indicate standard deviation.

Axial load to failure

There was no significant difference in axial load to failure between long nails and short nails overall or when subclassified into A1 and A3 fracture pattern groups. There was, however, significantly higher axial load to failure for A3 fractures (4970.3 N) when compared with A1 fractures overall (3073.0 N) (P < 0.0001) (Fig. 5).

Figure 5

Mean axial load to failure values. Horizontal bars at the top indicate pairs of constructs with significantly different values (P < .05). Error bars indicate standard deviation.

Mode of failure

All but 2 short nail constructs failed by screw cut out. The remaining 2 failed by failure at the lateral wall, both of which were in the A3 fracture group. All but 1 construct in the long nail A1 fracture pattern group failed by screw cut out. The remaining one failed at the distal interlock. The long nail A3 fracture pattern group had 3 constructs that failed by screw cut out, 3 that failed at the distal interlock, and 2 that failed at the lateral wall.

Discussion

Overall this study demonstrated no significant differences in axial load to failure values between long and short nail constructs in either A1 or A3 intertrochanteric fracture patterns in a composite femur model with low cancellous density and thin-walled cortical shell intended to mimic osteoporotic bone. Compared with long nail constructs, short nail constructs had significantly higher axial stiffness in the A1 fracture pattern and higher torsional stiffness in the A3 fracture pattern. These findings suggest that either long or short nail constructs could be appropriately utilized for management of intertrochanteric fractures in the elderly, including unstable fractures with posteromedial comminution or reverse obliquity patterns.

Based on axial and torsional stiffness testing in this study, short and long nail constructs exhibited similar stiffness for A1 or A3 unstable intertrochanteric fracture patterns when using a composite femur model. As discussed above, short nails demonstrated higher axial stiffness in A1 fracture patterns and higher torsional stiffness in A3 fracture patterns. We hypothesize that this may be due in part to the shorter working length offered by the more proximal interlock of the short nail construct. Additionally, the more proximal interlock screw in the short nail constructs engaged cortical bone, which may have provided more stability compared with the more distal interlock of the long nail constructs which engaged metaphyseal bone. However, it should be noted that some newer generations of long intramedullary nails have screw configurations which would enable diaphyseal engagement with the distal interlocking screw and could potentially affect the mode of failure. Notably, there was no significant difference in axial load to failure values between short and long nail constructs in either fracture group.

There are a number of reasons why treating unstable IT fractures with short rather than long IMNs may confer some clinical benefit to the patient. Long IMN insertion has been shown to require increased operative time compared with short IMN.[ This is likely due to the need for free-hand placement of distal interlocking screws.[ Furthermore, long IMNs have also been shown to result in increased blood loss and transfusion rates[ due to increased operative times and reaming, although reaming is not always performed. When reaming is performed, it is more frequently done for long nail constructs. In a systematic review of the literature, operative time was reported to be 18.5 minutes shorter for short IMNs and average blood loss was reported to be 86.7 mL for short IMN compared with 135.2 mL for long IMN.[ Cost is an additional factor that should be considered in implant selection. The cost of short IMN at our institution is approximately two-thirds that of long IMN and decreased operative times for short IMN may further contribute to decreased costs associated with their use. One study reported an average cost of $62 per minute of operating room time in the US.[ Based on the systematic review by Dunn et al,[ an assumed 18.5 minute decrease in operative time for short IMN would therefore project an additional $1147 in savings per case when short IMN is utilized. Additionally, 1 study demonstrated length of stay for patients treated with long nails when compared with short nails was, on average, 2.2 days longer and the associated hospitalization costs for these patients are approximately $3482 higher than those treated with short nails.[

While evaluating the incidence of periprosthetic fractures in these constructs was not the purpose of this study, it is interesting to note that no short nail constructs failed at the distal interlock site. Historically, short IMN fixation has been associated with higher rates of periprosthetic fracture postoperatively.[ This association is largely attributed to older implant designs[ and is thought to occur secondary to the increased proximal diameter and valgus curve of the earlier nails. In addition, stiff locking screws in the diaphysis are thought to have generated a stress concentration near the distal aspect of the implant,[ predisposing them to fracture. The relative increased risk of periprosthetic fracture with short IMNs is dramatically improved in newer generation designs[ due to the use of less rigid titanium alloy, decreased proximal diameter, tapered distal ends, smaller locking screws, and a more anatomic nail curvature.[ Several studies have been performed comparing rates of periprosthetic fracture between short and long IMN fixation and reported rates ranging from 0% to 3.3% for short IMN and 0% to 1.5% for long IMN.[ However, many of these studies have been underpowered[ and are variable in terms of types of implants and fractures included. One recent retrospective study demonstrated slightly higher incidence of ipsilateral refracture rate in patients managed with short nails when compared with long nails at 1, 2, and 5 years postoperative, but this did not reach statistical significance.[ In a recent systematic review that excluded the older generations of IMN, it was found that short nails had a refracture rate of 1.6% and all-cause reoperation rate of 5.02% whereas long nails had a refracture rate of 0.95% and reoperation rate of 3.82%. However, these differences were statistically insignificant.[ Other more recent studies corroborate these findings, showing long nails are not associated with lower rates of surgical failure and short and long nails have similar incidences of complications, readmissions, and mortality.[

Limitations of this study include the use of composite femur models instead of cadaver femurs since the ability of composite femur constructs to mimic the behavior of human tissue constructs is unknown,[ specifically in metaphyseal fracture models. Several studies have demonstrated that 4th generation sawbones whole bone properties are similar to that of cadaver tissue for properties such as stiffness, strain, fracture toughness, and fatigue crack propagation,[. Although this is a metaphyseal fracture pattern, and there are no known studies specifically validating sawbones models for proximal femoral metaphyseal fractures, other studies have validated the use of composite femur models in distal femoral metaphyseal fracture patterns demonstrating qualitatively reproducible results of 2 fixation constructs in cadaveric and composite femur models.[ Additionally, Zdero et al[ demonstrated no significant difference in pull out force or shear stress for cancellous screws inserted along the femoral neck axis between two-fourth generation composite femur and human cadaver femur models. Because 2 different fracture patterns were tested in this study, there would have been no inherent control in utilizing a pair of cadaver femurs to test the included constructs. Therefore, employing cadaver femurs as our model for this study would have introduced substantial variability amongst the specimens whereas sawbones composite femur models provide a standardized, consistent substrate[ to test with 2 different implants in 2 separate fracture patterns.

The fit of the nails within the diaphysis for the samples should be considered as well. Although we employed composite models with a capacious 13 mm canal and used a 13 mm reamer to standardize for surgeons who ream for both long and short nails, which is standard practice at some institutions, it should be noted that osteoporotic fractures frequently have more capacious canals than that. As a result, despite the fact that models with relatively low-density cancellous bone and thin-walled cortices were employed for this study, the constructs used in this study may not adequately reflect the fit of nails used for geriatric fractures. Conversely, a tighter fit at the distal end of the short nails likely creates a stress concentration in that area and given the historically higher rates of periprosthetic fracture with short nails, this model is relevant to consider when evaluating rates and modes of failure. It would be interesting and relevant to further investigate an additional comparison group with osteoporotic models and a more capacious canal.

An additional concern is the ability of this model to translate into clinical practice as the testing protocols may not accurately represent physiologic loading in terms of magnitude and direction of force. It is important to consider that IMN constructs typically do not fail in a single catastrophic load. Instead failures occur as a result of cyclic loading leading to fatigue failure. Therefore, our methodology and conclusions would have benefitted from a more rigorous cyclic loading test. Lastly, no lateral 3-point bending tests were performed which could have provided some further insight to the biomechanics of long and short nails under alternate loading patterns.

Overall, this study demonstrates that short nails are not biomechanically inferior to long nails in terms of axial and torsional stiffness or ultimate axial load to failure strength in unstable intertrochanteric fracture patterns in composite femur models with low-density cancellous bone and thin-walled cortices. These findings suggest either short or long nail constructs could potentially be safely employed for management of unstable intertrochanteric fractures in the geriatric population. Overall, A1 fractures appeared to be less stable with axial loading when compared with A3 fractures whereas A3 fractures were less stable with torsional loading. While this information may influence decision making regarding implant choice, this study supported the use of either a long or short IMN construct for either pattern. Further studies utilizing human cadaveric femora would be beneficial to further validate the results of this study. Additionally, prospective, randomized clinical trials in which constructs are put through more rigorous cyclic loading testing to evaluate stability and failures of long and short IMN constructs in unstable intertrochanteric fractures would be beneficial to confirm the findings of this study and translate these benchtop results into clinical practice.