Comparison of Screw Quantity and Placement of Metacarpal Fracture Fixation: A Biomechanical Study

Abstract

Background

It is recommended to have 6 bicortical screws for plate fixation of long bone fractures; however, many metacarpal fractures do not allow 6 screws due to size limitations and proximity of crucial anatomical structures. The purpose of this biomechanical study was to determine whether the mechanical properties of a 4-screw nonlocking construct are noninferior to those of a 6-screw nonlocking construct.

Methods

Metacarpal sawbones were used to simulate a midshaft, transverse fracture. Nonlocking bicortical screws were placed in the 6-hole plate, and the metacarpals were randomly assigned to 2 equal study groups: (1) 4 screws, 2 on either side of the fracture (4S); and (2) 6 screws, 3 on either side of the fracture (6S). The metacarpals were tested in a cyclic loading mode and load to failure in a cantilever bending mode.

Results

Maximum deflection was significantly higher for 4S compared with 6S. Cyclic root mean square (RMS) was also significantly greater for 4S at 70 and 100 N. There were no statistically significant differences observed between the 2 constructs for maximum bending load, bending stiffness, and cyclic RMS at 40 N. The maximum bending load in 4S and 6S was 245.6 ± 37.9 N and 230.8 ± 41.9 N, respectively; 4S was noninferior and not superior to 6S. Noninferiority testing was inconclusive for bending stiffness.

Conclusions

A 4-screw bicortical nonlocking construct is noninferior to a 6-screw bicortical nonlocking construct for fixation of metacarpal fractures, which may be advantageous to minimize disruption of soft tissues while maintaining sufficient construct stability.

Introduction

In the United States, there is an estimated total of 3 468 996 upper extremity injuries per year. ¹ Two of the most common types of upper extremity fractures are phalangeal and metacarpal fractures. ² Although most of the metacarpal fractures are treated nonsurgically, surgical stabilization is necessary for open fractures, unstable fractures, irreducible fractures, or high-energy impact injures. ³ Dorsal plate and screw fixation is a popular choice, with evidence that it is biomechanically superior to other methods such as Kirschner wires and simple lag screws.^4-7 The AO Foundation recommends a total of 6 bicortical screws, 3 proximal and 3 distal to the fracture site, for long bone fractures such as humerus, femur, and metacarpal.^8,9 However, many metacarpal fractures do not allow 6 screws of plate fixation due to size limitations, close proximity of crucial anatomical structures (tendons, muscles, and blood vessels), and irritation of the extensor tendons.^10,11 While increasing the number of screws or length of the plate increases the strength of fixation, these characteristics have also been correlated to increased healing complications. ⁹ Also, a fracture near the carpometacarpal or the metacarpophalangeal joint may only allow 2 screws as opposed to 3. Recent studies demonstrate that strategic placement of fewer screws can provide comparable fixation strength to 6-screw constructs and preserve desired clinical outcomes.^11-16 When varying the number of screws, 4 locked and 6 nonlocked bicortical screws demonstrated comparable fixation strength. ¹¹ However, it is not known how the fixation strength will compare between fracture fixations with different screw quantities when using the same plate and screw type.

The purpose of this biomechanical study was to further elucidate the optimal number of cortices needed for dorsal plate metacarpal fracture fixation—4 nonlocking bicortical screws versus the typical configuration of 6 nonlocking bicortical screws. The hypothesis is that there will be no significant difference in bending stiffness, maximum bending load, or maximum deflection between the 2 fixation constructs.

Materials and Methods

Fourth-generation composite corticocancellous biomechanical testing-grade third metacarpal sawbones (Sawbones Model 3416; Pacific Research Laboratories, Vashon, Washington) were selected for this study. A custom resin fixation cast was used to stabilize the sawbones during fixation. A midshaft, transverse fracture was simulated by cutting the sawbones at the center of their longitudinal axis with a bandsaw with a 1-mm blade. The plate was centered on the simulated fracture site. Nonlocking bicortical screws were placed in the 2.0-mm 6-hole/45-mm plate using the standard AO technique. The number of screws placed in each bone was determined by randomly allocating the metacarpals into 2 study groups: (1) group 4S: 4 screws, 2 on either side of the fracture; and (2) group 6S: 6 screws, 3 on either side of the fracture. A power analysis identified 25 samples necessary for bending and cyclic testing. The optimal sample size for each parameter was determined by iterating over the ranges determined from previous literature^{10,11,15,17-19} and selecting the sample size that would yield a significant difference for both parameters at a power of 0.95 for most of the iterations considered. Each group contained 13 metacarpals, with a total of 26 metacarpals used for the study. All screws were placed in plate holes closest to the fracture site, and screw lengths for each location on the metacarpal model were chosen to assure complete purchase of the far cortex. The metacarpals were then mounted in a custom polymethyl methacrylate fixture on the proximal end of the metacarpal, and cantilever bending tests were conducted on both groups using a material testing machine (Instron, Norwood, Massachusetts) (Figure 1). A perpendicular load was applied to a precise point on the distal end (4.5 cm from the proximal end/fixation block) using a wedge attached to the testing head of the machine. ¹⁹ Cyclic loading testing phases for the bone and plate construct were used to mimic the increased loading of the hand during a 6-week postoperative rehabilitation period of fracture healing.^19,20 Similar loading protocols have been used in previous literature.^21,22 Each specimen was first preloaded to 10 N and then loaded 1-40 N, 1-70 N, and 1-100 N in 3 successive phases. Each phase is representative of physiologic postoperative conditions: (1) active flexion with some resistance; (2) strong composite grasp; and (3) firm fingertip pinch.^17,20,23 Finally, those constructs that completed the cyclic loading protocol were monotonically loaded at a rate of 10mm/min until failure. Failure was defined as a sharp change in the bending load/deflection curve, and modes of fracture were the following: (1) screw pullout; or (2) fracture.^18,19 The failure mode was documented and photographed. For each construct, a load displacement curve was generated, and the yield point was recorded for comparison of tensile properties. Bending stiffness, maximum load, and mode of failure were determined for each construct. Bending stiffness was calculated from the linear portion of the load versus deflection curve. Root mean square (RMS) was calculated for the cyclic load testing.

Figure 1.

Setup for cantilever bending testing.

Statistical Methods

Means and standard deviations for bending stiffness, maximum bending, and maximum deflection were calculated for the groups. The assumption of normality and homogeneity of variance was tested using the Shapiro-Wilk test and the Levene test. Comparisons between the 2 constructs were carried out with the 2-tailed t test and the Mann-Whitney U test for parametric and nonparametric data, respectively. Differences were considered significant for P ≤ .05. If no statistical difference was found, we tested the hypothesis that the comparator group (4-screw construct) was noninferior to the 6-screw construct using a fixed 10% margin; the lower limit was used to define the margin.

Results

Descriptive statistics for each test are provided in Table 1. The maximum deflection was significantly higher (P = .009) for 4S (5.46 ± 2.05 mm) compared with 6S (3.80 ± 1.17 mm). The cyclic RMS was also significantly greater for 4S at 70 N (P = .003) and 100 N (P = .002). There were no statistically significant differences (P > .05) observed between the 2 constructs for maximum bending load, bending stiffness, and cyclic RMS at 40 N. Noninferiority testing was carried forward for each of the remaining comparisons.

Table 1.

Results of Cantilever Bending and Cyclic Loading Tests for 4-Screw and 6-Screw Groups (Mean ± SD).

Construct	Maximum bending load (N) a	Max deflection (mm) b	Bending stiffness (N/mm)	40 N cyclic RMS (mm) a	70 N cyclic RMS (mm) b	100 N cyclic RMS (mm) b
4-screw	245.6 ± 37.9	5.46 ± 2.05	47.0 ± 15.4	1.17 ± 0.76	1.47 ± 0.90	1.81 ± 1.14
6-screw	230.8 ± 41.9	3.80 ± 1.17	60.2 ± 19.4	0.82 ± 0.41	1.38 ± 0.76	1.48 ± 0.77

Note. RMS = root mean square.

^aNoninferiority of 4S compared with 6S.

^bSignificant difference.

The maximum bending load in 4S and 6S was 245.6 ± 37.9 N and 230.8 ± 41.9 N, respectively; 4S was noninferior and not superior to 6S (Figure 2a). Noninferiority testing was inconclusive for bending stiffness (4 S: 47.0 ± 15.4 N/mm, 6S: 60.2 ± 19.4 N/mm) (Figure 2b). 4S (1.17 ± 0.76 mm) was noninferior and not superior to 6S (0.82 ± 0.41 mm) for cyclic testing at 40 N (Figure 2c).

Figure 2.

Noninferiority plots for (a) maximum bending load, (b) maximum stiffness, and (c) 40 N cyclic RMS.

Note. RMS = root mean square.

All the constructs failed by fracture through the metacarpal (Table 2 and Figure 3). Of the 4S group, 54% (7 of 13) failed with screw pull-out at the proximal screw hole furthest from the applied force, and 46% (6 of 13) failed at the proximal bone-cement interface. Of the 6S group, 46% (6 of 13) failed with screw pull-out at the proximal screw hole furthest from the applied force, and 54% (7 of 13) failed at the proximal bone-cement interface.

Table 2.

Failure Mode of Constructs.

Failure mode	4 screws (n = 13)	6 screws (n = 13)
Bone-cement interface (mode I)	6	7
Most proximal screw hole (mode II)	7	6

Figure 3.

Examples of the 2 failure modes observed. Mode I describes failure at the bone-cement interface (left), and mode II describes failure at the most proximal screw hole (right).

Discussion

This study demonstrates that simulated transverse metacarpal fractures using 4 bicortical screws are noninferior to those repaired metacarpals with 6 bicortical screws for maximum bending load, maximum deflection, and RMS deflection. The deflection between the 2 groups remains similar at lower loads, but the 4S group experienced higher deflection as load increases. These findings suggest that although the 4S construct demonstrates more movement under load, it has similar maximum load and bending stiffness. Given that it is often impractical to place 6 cortices of fixation in the hands due to lack of space, it may be advantageous to use 4 screws to minimize disruption of soft tissues while maintaining sufficient construct stability.

The mean maximum bending load values for both constructs were several magnitudes greater than forces generated across metacarpals in vivo. Prior studies have shown that nearly all activities of daily living place less than 40 N of force on the metacarpal flexor tendons.^7,24-26 Given that the loads observed in this study are more than 20 times greater, it is unlikely that loads of this magnitude would be experienced during a normal postoperative period. Tannenbaum et al ¹⁹ showed that a cyclic loading model has the capacity to better represent acute rehabilitation compared with a sole load-to-failure model due to the importance of fixation fatigue failure. Our results show that the 4S and 6S constructs exhibit similar properties at physiologic loads, and therefore, the 4S construct demonstrates properties suitable for fixation and clinical rehabilitation.

Barr et al ¹¹ conducted a similar study and observed a maximum bending load of 185 ± 22 N and bending stiffness of 5.5 ± 0.5 N/mm for a 6-screw bicortical nonlocking construct in sawbones. These values are less than the values observed in this study. These findings may be attributed to Barr et al simulating a comminuted fracture by creating 3-mm diaphyseal gap, whereas this study used a completely reduced transverse fracture model. A gap is commonly used in biomechanical studies to isolate testing of the implant; however, this study opted for the transverse model to better replicate treatment of an acute noncomminuted fracture, a more common clinical scenario than nonunion.

This study compared the properties of 2 different nonlocking screw constructs. To our knowledge, most studies to date have only compared nonlocking with locking screws with less total fixation or compared different locking screw configurations. Ochman et al ¹⁶ showed that there was no significant difference between nonlocking and locking with bicortical fixation in metacarpal fixation. Barr et al ¹¹ supported this finding by demonstrating equivalent bending stiffness, torsional stiffness, maximum bending load, and maximum torque between constructs with either 4 or 6 bicortical locking screws. It should be noted that fewer than 6 screws have been evaluated in other long bone models. Lindvall and Sagi ¹² showed that diaphyseal forearm fractures treated with open reduction and internal fixation have similar complication (refracture, infections, nonunion) rates between 4 and 6 bicortical screws. Hak et al ¹³ also observed similar biomechanical behavior in a humerus fracture model of 4 versus 6 locking screws.

A number of factors (ie, trauma type, nutritional status, etc) influence fracture repair, but there is a consensus that the appropriate mechanical environment of the fracture site is essential for fracture healing.^27-35 Of the mechanical properties, plate axial stiffness is the dominant variable as it directly affects fracture healing, including vascularization and tissue differentiation.^29,33-35 A certain degree of rigidity is desirable, but a construct that is too stiff can retard bone healing. Although construct stability (ie, the ability to maintain fixation under load) is necessary, achieving the stiffest possible fixation may be detrimental. In other words, fixation stability is not synonymous with stiffness. ³⁰ This is especially relevant to the increased interest in locking screws, in which recent studies suggest that locked plate constructs excessively suppress interfragmentary motion, leading to insufficient bone healing.^33,36

Foux et al ³² contributed to the idea that there is a “sweet spot” for construct rigidity; although fixation systems that exhibited the extremes of the stiffness-elasticity spectrum resulted in the shortest healing time, the best healing (although somewhat longer duration) resulted from intermediary axial flexibility. Smith-Adaline et al ²⁸ expounded upon this by demonstrating that controlled mechanical stimuli promote bone formation; compressive strains induce bone formation, and tensile strains promote endochondral ossification. Epari et al ³¹ performed in vitro analysis on external fixation type of sheep tibiae fractures and found that, although the fixator type with the highest sheer stiffness was optimal, the fixator with the highest axial stability did not result in the highest failure moment.

This study has several limitations. A sawbone model was used as opposed to cadaveric specimens. However, the sawbones possess validated corticocancellous properties that have been shown to provide a consistent test medium to adequately simulate human bone. This study only evaluated unidirectional bending motion (unlike the multidirectional motions experienced in vivo) and did not assess other mechanical components such as torsion or shear. It should not be interpreted that axial stability is the only mechanical component that determines healing outcomes. It is possible that the differences observed (or lack thereof) can be partly attributed to torsional forces. However, apex dorsal bending was chosen for this study given that transverse metacarpal fractures are primarily due to forces perpendicular to the longitudinal axis of the bone. Furthermore, previous authors have demonstrated that flexor tendons are the predominant force generators across metacarpals,^37,38 and long bone fracture healing is particularly sensitive to axial movement and fixation stability.^39,40 Another limitation of the study is that only transverse fractures with cortical opposition were tested; unstable fracture patterns (eg, oblique and comminuted) were not included and may have had very different results with lack of cortical contact. However, dorsal plating is a common technique opted for in transverse noncomminuted fractures, and it remains unknown whether 4 bicortical nonlocking screws are as effective as 6 screws.

In conclusion, this study demonstrates that a 4-screw bicortical nonlocking construct is noninferior to a 6-screw bicortical nonlocking construct for a simulated transverse metacarpal fracture fixation in cantilever bending. The 4S construct may provide sufficient mechanical integrity similar to the 6S construct for dorsal metacarpal plating. The clinical implication of this study is that, with use of fewer screws, there will likely be less disruption of surrounding neurovascular structures and reduced risk for complications. Also, nonlocking screw constructs may limit excessive suppression of interfragmentary motion and optimize bone healing. Further studies will be necessary to assess other mechanical properties, such as torsion and shear, and to better correlate the findings of this biomechanical study with clinical outcomes. Additional studies will more confidently evaluate whether the 4S construct can provide equivalent healing without compromising the mechanical integrity of metacarpal plating in patients.