A comparison of conventional versus locking plates in intraarticular calcaneus fractures: A biomechanical study in human cadavers

1. Introduction 

Several large comparative clinical studies (Barla et al., 2004, Buckley et al., 2002, Rammelt et al., 2003, Randle et al., 2000, Sanders et al., 1993, Zwipp et al., 2004) have shown good to excellent results in younger patients with surgical treatment of displaced intraarticular calcaneal fractures. Controversy still persists for the treatment of calcaneal fractures in the elderly (Buckley et al., 2002, Essex-Lopresti, 1993, Paley and Hall, 1993). Besides lower physical demand, many elderly patients have lower bone quality impairing implant fixation leading ultimately to failure (Buckley et al., 2002). In these cases locking plates might potentially improve the outcome (Zwipp et al., 2004). Richter et al. (2005) showed in a biomechanical study that locking plates provide a greater stability during cyclic loading than conventional plates. No statistical difference in failure tests between the two plates was found. However, artificial bones with a single fracture pattern by defined osteotomies of the os calcis were used. In the present study, different clinical fractures of the os calcis on osteoporotic lower limb cadavers were produced through a direct impact to the specimens. Superior fixation strength of the locking plate would encourage the use of this plate in active elderly patients.

2. Methods 

Ten pairs of osteoporotic human cadaveric fresh frozen limbs cadavers were retrieved within 24h post mortem (median age 67, standard deviation 5.1; 3 males and 7 females). Radiographs of the foot were taken to exclude any osseous pathology of pre-existing disease or trauma. The bone mineral density (BMD) of the cadavers’ femoral neck and Ward’s triangle region was determined using dual-energy X-ray absorptiometry scans using a Hologic QDR-4500A densitometer (Hologic Inc, Waltham, MA, USA) with the running software v9.8D. Bone density was evaluated according to the WHO guidelines for osteoporosis and osteopenia (2.5 and 1 standard deviations respectively below the mean of a young healthy reference population of the same gender). To evaluate bone quality of the calcaneus bone density was also measured in the three locations intended for screw fixation using different software (Rat Whole Body v8.26a) (Fig. 1). The specimens were then wrapped in saline soaked linen, and stored at −80°C in sealed bags.

View Large Image Download to PowerPoint

Fig. 1. Three regions of interest were defined (R1–R3) and BMD was measured in each.

The intraarticular fractures of the calcaneus were created according to Carr et al. (1997) by dropping a weight of 148N guided by a steel rod which was inserted through a reamed (11mm) canal in the tibia 20cm above the ankle joint. The tibia was in an upright position and the foot in a 10° dorsally extended position supported by a frame to prevent it from sliding off the stand. A small notch was made in the sinus tarsi to produce a standard fracture pattern. The dropping height was lowered from 106cm (maximal available energy 156Nm) to 80cm (118Nm) after the first two specimens from different cadavers showed fractures of the distal tibia as well as through the talar neck. In spite of this modification one specimen sustained an unintended talus fracture and was also excluded from the study. Hence 14 paired specimens from seven cadavers remained in the study with isolated fractures of the calcaneus.

Two independent musculoskeletal radiologists then classified the fractures according to Sanders et al. (1993) using CT scans. The Böhler angles (BA) on lateral radiographs were also measured before and after fixation of the fracture and again after failure testing.

By means of an extended lateral approach the calcaneal fractures were reduced to less than 2mm of intraarticular displacement and stabilized with implants in a standard manner as described by Sanders and Gregory (1995). In both groups a total of nine bicortical screws were placed in the calcaneal bone: three beneath the posterior facet, three into the tuberosity fragment far dorsally, and three into the fragment of the anterior process close to the calcaneo-cuboidal joint (Fig. 3). All screws were inserted through the plate holes. The implants were either a locking or a non-locking stainless steel AO “Sanders plate” with self-tapping screws (Synthes, Sydney, Australia). In the calcaneal locking plate the conical threaded head screws are firmly locked into the also conical threaded plate hole providing axial and angular stability of the screw relative to the plate. In the calcaneal conventional plate the screws are tightened to compress the plate onto the bone. The actual stability results from friction between the plate and the bone (Fig. 2). The type of plate was randomly assigned to the right foot in each pair by drawing lots from the randomization envelope. The left foot was then stabilized with the other type of plate.

View Large Image Download to PowerPoint

Fig. 2. Calcaneal locking plate (left) and calcaneal conventional plate (right).

To minimize the soft tissue contributions to irreversible deformation during testing and to prevent the foot from slipping the heel pad was removed. The plantigrade hind foot was placed into a box containing dental cement (Glastone 3000, Die Stone, Type V, Densply, Perth, Australia). The box was then clamped to the baseplate of the material testing machine (Zwick Z010, Zwick Inc., Ulm, Germany) allowing free rotation. The tibia was oriented in line with the loading axis of the testing machine and proximally rigidly fixed in a custom made device (Fig. 3). The testing machine was controlled by special software (Test-expert; Zwick Inc., Ulm, Germany) on a standard IBM compatible personal computer that also directly stored the data.

View Large Image Download to PowerPoint

Fig. 3. Material testing setup using the Zwick Machine.

Each specimen was cyclically loaded 1000 times from a preload of 20N to a compressive load of 200N at a rate of six cycles per minute. The irreversible deformation was measured with the linear variable differential transformer on the Zwick Machine. The load to failure was then recorded at an actuator speed rate of 45mm/min. Ultimate load, ultimate deformation, and work to failure were recorded at the failure point, which was defined as the yield point on the load–deformation curve. Because the specimens were tested as an osteo-cartilago-ligamentous unit, the measured values reflect load, displacement and work to failure of the construct as a whole. After failure we dissected each specimen to identify the mode of failure.

3. Statistical analysis 

The Wilcoxon signed rank test was used to compare the results of the irreversible deformation testing, the failure testing and, the BMD measurements of the two groups. To compare the BMD measurements of the different areas of the calcaneus the Kruskal–Wallis test was applied. The Spearman’s correlation coefficient was used to assess the degree of correlation between values obtained by mechanical testing and BMD of different areas. Kappa statistics were used to analyse the interobserver variability of the CT based fracture classification. The hypothesis was that both plates would perform similarly in cyclic loading and in failure testing. P-values of <0.05 were considered significant.

4. Results 

4.1. Imaging 

All specimens were osteopenic. The bone density of the calcaneus was similar between the two groups (for all regions P>0.75). The bone below the posterior facet (R2) showed the highest density, followed by the anterior process region (R1) and the calcaneal tuberosity (R3). Both groups displayed significant (P<0.05) differences between these three areas.

The distribution of fracture type for both groups was similar (Table 1). The interobserver variability of the classification system was 0.57 (95% confidence interval (CI) 0.46–0.68). The subclasses were then combined and assessed. An agreement between the general classes gave a kappa value of 0.61 (95% CI 0.52–0.70). The BA was restored within the 95% CI in 5/7 cases using the locking plate and in 6/7 cases using the conventional plate.

Table 1. Fracture classification based on CT scan

	Examiner	Locking plate		Conventional plate
		One	Two	One	Two
	1	1			1
	2A	1	1	2
	2B		2
	2C			1	1
	3AB	1	2	1	2
	3AC	1	1		1
	3BC
	4	3	1	3	2

4.2. Mechanical testing 

After 1000 cycles, the mean irreversible deformation excluding the first cycle was 2.75mm (SD 0.47) and 3.49mm (SD 0.96) for the locking and conventional plates, respectively (Fig. 4, Fig. 5) (P=0.031).

View Large Image Download to PowerPoint

Fig. 4. Irreversible deformation (y; excluding the first cycle) during the cyclic loading sequence, for plates with (a) and without locking screws (b).

View Large Image Download to PowerPoint

Fig. 5. Ultimate load, ultimate deformation and, irreversible deformation for each group.

The ultimate load to failure, the ultimate displacement, and work to failure for the locking plates were 3818N (SD 1712), 15.9mm (SD 4.5), 23.1J (SD 13.8) and for the conventional plates 3176N (SD 1424), 17.3mm SD (5.6), and 29.3J (SD 23.2), respectively (Fig. 5). In all cases the locking plates showed a higher load to failure. In 2 constructs a higher ultimate displacement and, in 1 construct a higher work to failure were recorded. A significant difference between the two groups could be found for the load to failure (P=0.031) but not for the ultimate displacement and the work to failure.

A low bone mineral content in the area of the posterior facet correlated significantly with an increased irreversible deformation for conventional plates (r=0.87, P=0.024) but not for locking plates (r=0.36, P=0.29).

At the point of failure all specimens failed by loss of reduction. The force applied to the tibia caused subsidence of the posterior facet in both groups. The locking plate failed by bending of the plate itself while the screws maintained their plate-screw angle, whereas in conventional plates the plate remained intact but the screws angulated relative to the plate by pivoting in the plate hole.

5. Discussion 

The optimal treatment of displaced intraarticular calcaneal fractures in the elderly patient is challenging. The suitability for internal fixation in this population group should be determined by the patient’s overall condition and functional demand rather than their age (Rammelt and Zwipp, 2004). Herscovici et al. (2005) reported good clinical results in elderly patients following an operative treatment. He stated that careful patient selection is necessary as individuals presenting with severe osteopenia have the risk of implant loosening with loss of reduction. In patients with poor bone quality locking plates have been shown to have an advantage over conventional plates (Gautier and Sommer, 2003, O’Driscoll, 2005, Ring et al., 2004, Stoffel et al., 2003).

The present study investigated the fixation strength of locking versus non-locking screws in the fixation of intraarticular fractures of the calcaneus in osteoporotic human cadavers. The locking plates provided a significant increase in fixation strength with a lower irreversible deformation during cyclic loading and a higher load to failure. When comparing the ultimate deformation and the work to failure no statistical difference between the two groups could be found.

The mean age of our osteoporotic cadaveric specimens was 67 years with a low standard deviation (5.1), providing suitable material for demonstrating a difference in constructs stability between the two types of plate fixation. Furthermore randomizing within the specimen pairs for the alternate forms of fixation helped reduce the effect of differences in bone quality and made the data more comparable. We used the dynamic single impact fracture model described by Carr et al. (1997). In order to avoid additional fractures around the ankle joint we had to lower the dropping height from 106cm to 80cm. With this slight modification we produced clinically relevant fracture patterns with variation of the primary fracture line and different patterns of comminution of the posterior facet (Buckley et al., 2002, Rammelt et al., 2001, Sanders, 2000). These variations cannot be seen by creating intraarticular “fractures” by defined osteotomies (Badet et al., 1999, Lin et al., 1998, Richter et al., 2005). In spite of only fair agreement among the two radiologists in classifying the fractures, we consider the fractures as being consistent with typical fractures encountered clinically in impact trauma. We encountered low interobserver reliability in classifying our fractures but consider this to be a common problem of classifications in general and not unique to the Sanders classification (Bhattacharya et al., 2005).

The rate (six cycles per minute) and the load (200N) used for cyclic compressive loading mimics crutch walking as described in earlier biomechanical studies (Chapman et al., 1996, Lange et al., 1990). Others used compressive loads during cyclic testing of either 100N (Carr et al., 1997) or 800N (Richter et al., 2005) or tested only single load to failure (Lin et al., 1998).

During cyclic loading the resistance to irreversible deformation was significantly higher in locking plates compared to conventional plates. This supports the findings of Richter et al. (2005). In our model the deformation was most likely due to a subsidence of the posterior facet. We measured the displacement only in the primary loading axis with the setup for both sides and plates being identical. Only in the conventional plate group could a significant correlation between an increased irreversible deformation and a low bone mineral density be found. Accordingly it can be assumed that the irreversible deformation of a construct with a locking plate is mineral content independent.

In all cases the load to failure was well above the average physiological force of 755N acting on the calcaneus at heel strike during normal gait (Hutton et al., 1982). The load to failure in our series was relatively high in both groups, which might be related to the adequate restoration of the Böhler’s angle (tripod effect) as described by Letournel (1993). However, the results of failure testing were significantly higher in the locking plate group and did not correlate with the fracture type. From a failure point of view the success of surgery is mainly dependent on the bone quality and hereby indirectly, also the implant design (locking vs. conventional) and the restoration of the Böhler’s angle. This is supported by the fact that the restoration of Böhler’s angle has been found to be crucial for good long term results independent of the treatment mode (Paul et al., 2004). The low density of the tuberosity (R3) was not reflected in the mechanical strength of the different implants or the failure mode. All constructs failed in the area of the main direct axial load impact, the posterior facet. No failure was detected from the loss of fixation of the anterior or posterior process in any of the implants. These findings, together with the findings from the cyclic loading tests, suggest that a plate with locking holes beneath the posterior facet and compression holes in the rest of the plate might provide as stable a construct as a plate with all locking screws. The fact that Richter et al. (2005) found no significant difference in the load to failure might be related to the specimens tested. They tested composite bones with the mechanical properties of a young patient. It has been shown that in good bone quality locking plates demonstrate only a subtle mechanical superiority compared to compression plates (Gardner et al., 2005, Wagner, 2003). In addition to the bone density the failure mode must be attributed to the fixation principle (locking vs. conventional plate) since the plate thickness between the screw holes is the same for both. Also with the screws in place the construct thickness is the same since the locking screws are countersunk into the plate. As a result of these facts the locking and the conventional plate are of equally low profile and reduce the risk of a possible wound breakdown due to stretching the overlying soft tissue over the implant as requested by Carr et al. (1997).

The absence of stabilizing forces of the muscles in our cadaveric limbs compared to that in vivo is a limitation of the study. Further limitations include the uniaxial mode of loading and the measurement of the displacement of the entire construct registered by the material testing machine. However, it has been shown that the motion measurements of the mechanical testing machine are equivalent to those recorded when measuring the motion between the plate and the bone (Richter et al., 2005). In our experimental setting we used either conventional or locking screws. In a clinical setting however, the surgeon may chose to use compression screws separate from the plate to achieve interfragmentary compression followed by a locking plate to stabilize the fracture. This might help to maintain the reduction as well as to improve the fixation strength.

This is the first biomechanical study performed on fresh frozen human cadavers that investigated the fixation strength of plates with and without locking screws in intraarticular calcaneal fractures. We found significantly less irreversible deformation regardless of bone quality and a significant higher load to failure in locking plates in osteoporotic bone. We conclude that locking plates are beneficial when treating displaced intraarticular calcaneal fractures in patients with osteoporotic bone.