In-vivo stiffness assessment of distal femur fracture locked plating constructs

Abstract

Background:

The purpose of this study was to design and validate a novel stiffness-measuring device using locked plating of distal femur fractures as a model.

Methods:

All patients underwent a laterally-based approach, with a bridging locked construct after indirect reduction. A custom and calibrated intraoperative stiffness device was applied and the stiffness of the construct was blindly recorded. Fourteen of twenty-seven patients enrolled with distal femur fractures (AO/OTA 33A and 33C) completed the study. Correlations between stiffness and callus formation, working length, working length/plate length ratio, number of distal locking screws, and fracture pattern were explored.

Findings:

Callus and modified radiographic union scale in tibias scores as a linear function of stiffness did not correlate (R2 = 0.06 and 0.07, respectively). Construct working length and working length to plate length ratio did not correlate to stiffness (R2 = 0.18 and 0.16 respectively). A combined delayed and nonunion rate was 14%. Lower extremity measure scores were not statistically different when comparing delayed and nonunion with healed fractures.

Interpretation:

The lack of correlation may have been due to the mechanical properties of the plate itself and its large contribution to the overall stiffness of the construct. To our knowledge, clinically relevant stiffness has not been described and this study may provide some estimates. This methodology and these preliminary findings may lay the groundwork for further investigations into this prevalent clinical problem. Other parameters not investigated may play a key role such as body mass index and bone mineral density.

INTRODUCTION

Delayed healing or nonunion occurs in an estimated 5–10% of all fractures. Animal and computational studies have demonstrated that the ideal fixation stiffness occurs within a range to allow fracture healing.^1–3 Excessive stiffness of bone-plate constructs may be responsible for delayed union or nonunion in the treatment of comminuted distal femur fractures.^{4, 5} Biomechanical studies have shown that plate-screw construct variables can affect stiffness, such as the plate working length, plate length, screw length and number, and using dynamic locked plating.^6–9 Currently, there is no way to objectively assess the stiffness of constructs applied intraoperatively. The exact appropriate stiffness range is unknown. To address this issue, a novel device to intraoperatively quantify the stiffness of locked plating of distal femur fractures was designed and validated. In-vivo stiffness has not previously been quantified intraoperatively. The rationale for this study was to examine and quantify variables of a bone-plate construct for the treatment of distal femur fractures that correlate to the measured stiffness and potentially callus formation. This would be clinically relevant as surgeons could modify the construct variables at the time of application to maximize union rates. The purpose of this study was to design and utilize a novel stiffness-measuring device in the treatment of distal femur fractures fixed with locked plating. To our knowledge at the time of this manuscript no device existed. We hypothesized that a positive correlation would exist between stiffness and callus formation, working length (WL), working length plate length ratio (WL/PL), and the number of distal locking screws.

METHODS

Design and Validation of Custom Stiffness Measurement Device

We designed, built, and validated a custom device (Figure 1–3) to measure stiffness of a distal femoral fracture plate (LISS, Less Invasive Stabilization System, Synthes). The device consists of a screw- driven linear actuator that is manually controlled, in series with a force transducer (75 N, Omega) connected to a data acquisition computer. At each end of the device are two ball joint rod ends that attach to posts or drill guides, which attach at their other ends to the fracture plate. Linear extension is produced in fixed increments by rotating the actuator head in a stepwise manner. Extension of the device produces varus bending of the plate given that the plate is placed laterally on the femur. We validated the stiffness measurement device by comparison to a materials testing machine (Instron 5866). Extension tests were performed on plate-femur constructs using the device and then repeated using the Instron. All tests were run in triplicate. The plate was a stainless steel, 13-hole LISS distal femur plate. The “femur” was a synthetic model femur (Sawbones 4^th gen. composite femur, Pacific Research) with a 1 cm segment removed to simulate a comminuted metaphyseal fracture without cortical contact. A set of three femurs was used, each in three configurations (n = 9 test constructs). The configurations were selected to produce a range of construct stiffness by varying placement of the distal screw in the proximal fragment (hole 2 vs. 6) or the type of screw fixation (standard bicortical vs. near-cortical overdrilled locking). Stiffness was determined from the slope of the force-extension curve, seen in graph from one sample test (Figure 4). We evaluated stiffness of the same nine test constructs in axial compression, in order to approximate single-legged stance loading. A vertical force was applied through the femoral head; the femoral condyles were supported in a molded socket. Two pairs of reflective markers were placed on either side of the fracture gap, and their motion was recorded optically (Qualisys). Near and far cortical displacements were computed at 500 N force (~0.75 BW, 70 kg subject). The device was later modified and validity confirmed for the 4.5mm VA-LCP curved condylar plate (Synthes, West Chester PA), which was utilized in the clinical portion of the study. The number of sample femurs was later extended an additional 24 yielding a total sample size of 33. The sample size calculation for the analysis of reliability was performed following the procedures in Walter et al.¹⁰ Acceptable reliability for this study was set at 0.80, with an alpha level of 0.05 and a 0.20 Type II error.¹⁰ Assuming that the true reliability of the novel device is 0.90, a minimum sample size of 33 specimens sampled three times was required. Preliminary stiffness data from the nine specimens indicated an average between-device difference of 0.011 N/mm (SD=O.105), With this variability and a total sample size of at least 33, this experiment had power of 80% to detect a between-device difference of at least 0.05 (alpha=0.05, two-tailed paired t-test). Stiffness was determined from the slope of the force-extension curve. Repeatability was seen with linear force-displacement behavior. The device produced a repeatable stiffness value with an average CV of 1.7%. The intraclass correlation coefficient for the measurements of the device and the Instron was 0.93 (95% CI 0.8–0.98). The accuracy of the novel device was correlated to the Instron values during extension test with RMS error 0.1N/mm

Figure 1.

Photo of Custom Device Test Set Up

Figure 3.

Photo of Instron Compression Test Set-Up

Figure 4.

(A) Force-extension plots for a representative specimen measured using the custom device and using the Instron. The plots are nearly identical. (B). Stiffness determined using the Instron was strongly correlated with stiffness determined using the custom device.

Clinical Study

Patients who sustained distal femur fractures with the operative plan of locked distal femur plating were prospectively enrolled into the study from October 2012 to July 2015 from a level I trauma center. IRB approval was obtained prior to the study. Fixation constructs were defined as either bridging or absolute rigidity constructs. Absolute rigidity constructs were defined as any bone-plate construct with an anatomic reduction with lag screw(s) or compression across the major meta-diaphyseal fracture fragment to induce primary healing. Bridging constructs were defined as any bone-plate construct that was placed spanning fracture or comminution to induce secondary healing, typically following indirect fracture reduction. Patients were excluded if they did not agree to participate or if they were lost to follow-up for any reason during the study. Patients were excluded if the plate was not the 4.5mm VA-LCP distal femur plate (Synthes, West Chester, PA), or if the construct was designed to be an absolute rigidity construct. Fracture patterns were categorized by the AO/OTA classification system.¹¹ Patient demographics were collected including age, sex, BMI, tobacco use, diabetes, immunosuppression status, and mechanism and laterality of injury. Fourteen of the 27 enrolled patients completed the study met the inclusion criteria. Three additional patients completed the study but had absolute rigidity constructs with the intention of primary bone healing and were therefore excluded from the main analysis. These patients were later compared to the inclusion population for comparison between bridging and absolute rigidity constructs. Average age (N=14) was 67 years (range: 40–98) and average BMI was 31.9 (range: 18–47.6). Four of the 14 patients sustained injuries from a high-energy mechanism and the rest were ground level falls. No fractures were open. Ten of the fractures were classified as AO/OTA 33A and four as AO/OTA 33C. Four fractures in the cohort were above total knee arthroplasty (TKA), with fractures in the supracondylar region above the implant. The femoral components were stable.

Injury radiographs were examined by the lead author to determine the above fracture patterns. Intraoperative variables including number and type of screws used, plate length (PL), and working length (WL) were chosen at the discretion of the three orthopaedic traumatologists participating in the study. Working length was defined as the distance between the first proximal screw and the first distal screw from the fracture line(s), eg, the length of plate without screws spanning the fracture zone. All patients were treated with one of two lateral surgical approaches, either full exposure or a minimal and percutaneous method, also determined by the treating surgeon. Medial stability from the fracture pattern was not controlled in the study. Following application of the construct based on standard principles, the stiffness measuring device was applied to the implant and measurements were taken as described above (Figure 5). Implants and screws were not adjusted after the measurements were taken. After in vivo use, the device was calibrated by running the device on an original sawbone and plate construct to confirm the device was accurately reading stiffness.

Figure 5.

Novel stiffness measure device applied to distal femoral locking plate

All patients were followed clinically for one year, standard follow up 2 weeks, 6 weeks, 3,6, and 12 months, and data were collected including radiographic and clinical union, patient-reported functional outcomes Lower Extremity Measure (LEM) scale, fixation failure, postoperative infection, delayed or nonunion, and unplanned return to the operating room. Patients were kept non weightbearing for a minimum of 6 weeks, and progression of weightbearing was at the discretion of the treating surgeon. Clinical union was defined as pain free full weightbearing with radiographic evidence of bridging callus of fracture healing. Radiographic union was defined as mature callus formation seen on multiple cortices with orthogonal radiographs. Nonunion was defined using the FDA standards: the fracture not healed in 9 months or without radiographic signs of progression toward healing for 3 consecutive months. A callus score and a modified radiograph union scale in tibial fractures score (mRUST) were determined at the 3 month follow-up with orthogonal radiographs of the distal femur.¹² The scores were averaged between the 3 treating surgeons after the blinded radiographs were graded. The callus score used was based on the amount of callus noted on the 3 month radiographs and was based on a previously published scale (0 – no, 1 – minimal, 2 – moderate, 3 – robust).¹³

Analysis of the data collection was done using Excel (Microsoft, Redmond, WA). Linear regression with R² values were calculated between stiffness and the following: WL, WL/PL, mRUST, and callus score. A two-tailed student t-test was used to compare means of LEM, Callus score and mRUST score with an alpha of 0.05 being statistically significant. Groups included fracture patterns (A vs C), smokers and non-smokers, diabetics (DM) and non-diabetics, and patients who had delayed or nonunions to those who did not. An ANOVA analysis was used to evaluate if the number of distal screws used affected the measured stiffness. The 3 groups compared included 4 constructs with 4 screws, 4 constructs with 5 screws, and 6 constructs with 6 screws.

RESULTS

Construct stiffness did not strongly correlate with either callus score or mRUST score (R² = 0.06 and 0.07, respectively). There was no strong correlation between stiffness and WL or stiffness and WL/PL ratio (R ² = 0.18 and 0.16 respectively). Sub analysis of the A and C fracture pattern did show improvement in linear correlation with WL versus stiffness and WL/PL versus stiffness for C-types (R ² = 0.59 and 0.32 respectively) (Figure 6) but not for A –type (both R ² = 0.12). When stratified for the number of distal screws (4,5, or 6), stiffness was not statistically different (p = 0.887) based on the ANOVA analysis.

Figure 6.

WL and WL/PL versus Stiffness with Linear Regression for AO/OTA 33C fracture pattern.

Three patients had complications: one had a delayed union and two patients had an aseptic nonunion. There was no statistical difference in the stiffness measurements when comparing patients with a complication to patients without a complication, 4.49 N/mm versus 4.63 N/mmm, respectively and p =0.75. Mean LEM score for patients who had a complication (39) compared to no complication (61) was not statistically different (p = 0.17). There was no difference in stiffness or LEM when comparing patients who had a history of smoking to nonsmoker, DM to non-DM, and between A-type and C-type fractures (Table 1).

Table 1.

Comparison of outcomes and callus formation based on potential risk factors.

	LEM		Callus Score		mRUST Score
	Mean	p value	Mean	p value	Mean	p value
Smoker (N = 3)	58	0.83	1.8	0.74	8.6	0.74
Non Smoker ( N = 11)	51		1.5		8
DM ( N = 5)	45	0.25	1.5	0.76	7.5	0.27
Non DM ( N = 9)	60		1.6		8.4
Delayed Union/ Nonunion (N = 3)	39	0.17	1.5	0.11	7.3	0.11
Healed ( N = 11)	61		1.7		8.6
33A ( N = 10)	53	0.84	1.5	0.42	7.8	0.36
33C ( N =4)	57		1.8		8.8

Three patients that completed the study but were excluded in the above analysis were designed to have absolute stability and primary bone healing. There was no statistical difference in stiffness between the absolute rigidity constructs (mean stiffness of 4.79 N/mm [range: 1.07–7.67]) and the relative stability constructs stability (mean stiffness of 4.79 N/mm [1.76–8.20]), p = 0.99. The mean WL for the absolute and relative rigidity constructs were 79 mm and 90 mm respectively and they were not statistically different (p =0.57).

DISCUSSION AND CONCLUSIONS

Delayed unions and nonunions are still an important concern for surgeons treating all fractures. Recently, distal femur fractures have been studied frequently for this complication. Nonunion rates of 0 to 20% for distal femur treated with lateral locking plates have been reported.^14–19 Stiffness has been implicated for these relatively high nonunion rates, but no clear guidelines for the adjustment of stiffness exists. Rodriguez et al examined 283 patients and found a nonunion rate of just under 10%.¹⁷ Additionally, they found obesity, open fractures, infection, and stainless steel plates were independent risk factors.¹⁷ That same group recently published a study analyzing the mechanical construct characteristics and found the “rigidity score” to be of significance.¹⁸ However, the study utilized an arbitrary scale based on implant material, proximal screw density, and if screws crossed fracture lines, which is different than measuring actual construct stiffness. The use of scores such as this led the authors to explore the concept of measuring actual construct stiffness. The concern of overly stiff constructs had lead researchers and surgeons to utilizing more flexible implants (titanium vs. steel), advanced techniques (over drill of near cortex with locking screw), and new screw designs such as far cortical locking screws^{20, 21}. These attempts at decreasing construct stiffness are arbitrary and there is no objective data guiding the optimal amount of stiffness needed to ensure union. In this small prospective case series, the combined delayed and nonunion rate was 14%, which was within the literature range. There was no identifiable risk factor to affect the LEM score in this series, including delayed union or nonunion, smoking status, history of diabetes, and patients with intra-articular injury compared to supra-condylar injuries.

In this study, we did not find correlations between callus formation or healing, and construct stiffness. We also did not find correlations between callus formation and WL or WL/PL. This may have been due to the mechanical properties of the plate itself and its large contribution to the overall stiffness of the construct. A large prognostic study evaluating risk factors for nonunion or delayed union in distal femur fractures found that plate length (< 8 hole) did influence this risk.¹⁵ In our series all plates were between 14 and 16 holes in length. In fact 15 out of 18 of the plates used were 14 hole-plates. Had more plates of varying lengths been used the results may have been different.

There were other limitations to this study. First, this was a small series. Seventeen patients completed the study, fourteen of whom had relative stability constructs, although a prospective power analysis was done during the lab and pilot portion of the device. Additionally, this study only evaluated a single implant with a single material (stainless steel), so our results might not extrapolate to all implants. A particular strength of the study was the novel method to evaluate stiffness of a completed construct intraoperatively. The device was validated in the lab to be accurate before being utilized in the in-vivo study. The study was prospective with clinical outcomes as well. The device was calibrated between each use to minimize faulty stiffness measurements, as described above.

Interestingly, three patients who completed the study were excluded based on the construct design. Those patients with absolute rigidity constructs had similar stiffness to the bridged constructs (overall and for 33C subtype). Additionally, 2 of the 3 patients appeared to form callus based on the callus and mRUST scores, which one would not expect. However both of these injuries were open, so it is difficult to determine if the callus seen was actually heterotopic ossification (HO) from extensive muscle injury (one patient had secondary surgery for excision).

To our knowledge, no previous attempt has been made to intraoperatively measure construct stiffness in a prospective clinical series to determine correlations to healing. The concept behind this study was that it may be possible for a surgeon to measure intra-operative stiffness, and once the ideal stiffness range was determined with observational studies, the construct could be modified accordingly. Clearly, this would provide powerful real time information for a surgeon to reduce the risk of delayed or nonunion. A power analysis was unable to be performed due to the lack of knowledge of clinically relevant stiffness, though this study may provide future studies with stiffness estimates. This methodology and these preliminary findings may lay the groundwork for further investigations into this prevalent clinical problem and allow for future investigation into other relevant factors such as bone mineral density, and body mass index.

Figure 2.

Photo of Instron Tension Test Set-Up

Highlights.

• A custom device was used to measure stiffness of distal femur fixation constructs

• This study provides estimates for in-vivo plate stiffness for fracture constructs

• Nonunions remain a prevalent problem; this methodology warrants further study