Fully threaded sacroiliac lag screws have higher load to failure when compared to partially threaded screws: A biomechanical study

Abstract

The purpose of this study is to compare biomechanical properties of fully and partially threaded iliosacral screws. We hypothesise that fully threaded screws will have a higher yield force, and less deformation than partially threaded screws following axial loading. Twenty sawbone blocks were uniformly divided to simulate vertical sacral fractures. Ten blocks were affixed with fully threaded iliosacral screws in an over-drilled, lag-by-technique fashion whilst the remaining ten were fixed with partially threaded lag-by-design screws. All screws measured 7.3-mm x 145 mm, and were inserted to a 70% of calculated maximal insertional torque, ensuring uniform screw placement throughout across models. Continuous axial loads were applied to 3 constructs of each type to failure to determine baseline characteristics. Five hundred loading cycles of 500 N at 1 Hz were applied to 4 constructs of each type, and then axially loaded to failure. Force displacement curves, elastic, and plastic deformation of each construct was recorded. Fully threaded constructs had a 428% higher yield force, 61% higher stiffness, 125% higher ultimate force, and 66% lower yield deformation (p < 0.05). The average plastic deformation for partially threaded constructs was 336% higher than fully threaded constructs (p = 0.071), the final elastic deflection was 10% higher (p = 0.248), and the average total movement was 21% higher (p = 0.107). We conclude from this biomechanical study that fully threaded, lag-by-technique iliosacral screws can withstand significantly higher axial loads to failure than partially threaded screws. In addition, fully threaded screws trended towards exhibiting a significantly lower plastic deformation following cyclical loading.

1. Introduction

Pelvic ring fractures account for 3% of all skeletal fractures and can be associated with significant functional deficits if unstable injuries progress to malunion.¹ Surgical fixation may minimize the risk of long term pain and gait disturbance in unstable pelvic ring fractures.2, 3, 4 Percutaneous iliosacral (IS) screw fixation, typically with a partially threaded, cannulated screw, for stabilisation of the sacroiliac (SI) joint has been identified as a favourable method of posterior ring stabilisation and requires a minimally invasive approach.⁵

Although the utility of using a partially threaded screw arises from only requiring a single drill hole thereby eliminating the need to over-drill through the cancellous bone of the posterior superior iliac crest, partially threaded designs have been shown to have reduced initial screw stiffness and yield load when compared to fully threaded screws when used for fixation of ankle fractures.⁶ Furthermore, the incidence of IS screw loosening has been observed as at a rate as high as 17.3%, with up to 11.8% of patients requiring revision surgery.⁷

No studies to date have investigated or compared the biomechanical properties of a fully threaded against a partially threaded cannulated 7.3 mm core diameter screw, which is most commonly utilised in the reduction and fixation of sacroiliac joint disruption. The purpose of this biomechanical study is to compare the mechanical behavioural properties of fully and partially threaded iliosacral screws within a sawbones model simulating cancellous bone fixation, under maximal torque, axial load to failure, and cyclical axial loading. We hypothesise that a fully threaded lag by technique 7.3 mm screw will achieve a higher insertion torque, higher load to failure, and less displacement with cyclical loading when compared to a 7.3 mm partially threaded screw.

2. Methods

Ten 7.3  mm × 145  mm partially threaded screws (32 mm threads), and ten fully threaded 7.3  mm × 145  mm screws (model 209.745, 209.945; DePuy Synthes, Paoli PA) were obtained for analysis. Twenty sawbones blocks (model SKU:1522-02; Pacific Research Laboratories, WA) were cut to the dimensions of 25 mm and 155 mm, for testing. The constructs that were created were not intended to be representative of the sacro-iliac joint given the inability to simulate the surrounding soft tissues and ligaments. We opted against using a sawbones pelvis model for this same reason. The primary goal of our construct was to create a sawbones model, affixed with long sacro-iliac screws, that were capable of undergoing stable axial testing, in order to investigate their specific subsequent biomechanical properties. A 15pcf density was utilised in concordance with the American Society for Testing and Materials International standard for biomechanical testing of osseous screws in cancellous bone. As our sawbones did not simulate a cortical bone surface, we created a washer using 3/16 inch sheet aluminium to prevent the screws from subsiding into the sawbones during insertion. We also recognise that there was not simulation of cortical bone within the shell of our sawbones block, however our study intent was to calculate the properties of screws that had purchase within cancellous bone, rather than a screw that had multi-cortical purchase.

Screws were placed by a single surgeon to limit variation in technique. A 2.8 mm threaded guidewire was inserted into the sawbones blocks in perpendicular orientation to the simulated SI joint. A 5.0 mm pilot hole was drilled across the SI joint for both fully threaded and partially threaded screws. An overdrilled glidehole of 7.3 mm was created for the fully threaded screw samples up to the level of the SI joint. The screw was inserted using a cannulated hexagonal screwdriver over the guide wire, along with our manufactured and standard washers. The guidewire was subsequently removed prior to biomechanical testing (Fig. 1).

Fig. 1.

Photograph demonstrating methodology used to obtain maximum insertional torque readings. Clamp apparatus can be seen that was utilised to hold construct to base and prevent rotation on baseplate. Cannulated screwdriver power attachment was extended from MTS and utilised over guidewire to obtain screw specific torque insertion. Screwdriver power attachment was substituted with regular manual cannulated screwdriver when constructs were created B) Photograph demonstrating apparatus and construct setup, with fabricated and manufactured washer prior to mechanical testing. A cage was constructed and utilised to prevent anterior and posterior tilting during testing, and retain the axial force vector applied by MTS.

Biomechanical testing was performed using an MTS 858 Mini Bionix (MTS Systems Corporation, Minneapolis MN).

2.1. Torque

An insertion torque of approximately 70% of maximum is recommended to prevent screws from stripping during insertion.^8,9 In order to provide a standardised insertion torque for axial load testing, we performed a maximal insertion torque (MIT) to failure analysis using one screw of each type. The MTS was utilised to insert the screw whilst torque readings (N-m) were continuously obtained. MIT was defined as the maximum torque reading obtained prior to failure, which was defined as the highest achievable torque prior to a reduction in torque readings, which was assumed to be secondary to the screw stripping. Following obtention of this value, all subsequent constructs were inserted at 70% of the MIT for the designated screw.

To simulate an environment similar to that in the operating room, we proceeded to insert the same screws manually to these designated values. The construct was clamped to the baseplate of the MTS and the screw was inserted manually with a regular cannulated screw driver, which allowed us to obtain continuous torque readings. We found that it was impossible to insert the fully threaded screw to the insertional torque previously designated by the MTS, as the magnitude of torque was too high. As a result, we opted to utilise the largest, most reproducible torque value of 6.58N-m for the fully threaded screws.

2.2. Load to failure

To determine baseline characteristics of each of the screw types, an axial load to failure was performed for 3 screws of each type. Constructs were randomised to a consecutive order for testing to minimize consistency bias using MATLAB® (MathWorks). An axial shearing load was applied perpendicular to the construct at a rate of 0.25 mm/s, until failure was achieved. Failure was defined as a maximum displacement of 3.0 cm, as this was felt to be the largest displacement that would be observed prior to the construct physically breaking, and would take into account the compression of the testing material itself.

Force displacement graphs were obtained for each tested construct. Stiffness (N/mm) was calculated by calculating the gradient of the initial linear region of each plot. The yield force (N) was obtained by identifying the force value at the intersection of the plot and a 0.05 mm parallel offset line to the initial linear region. Ultimate force (N) was defined as the maximum force obtained at any point during testing, prior to failure. Yield deflection (mm) was defined as the total deformation of the screw, at the point prior to yielding.

2.3. Cyclical loading

Cyclical load testing was performed for four screws of each type. Again, constructs were randomised to a consecutive order for testing. A 500 N axial load was applied for 500 cycles at a frequency of 1Hz.^10,11 Hysteresis curves of force against displacement were obtained. Plastic deformation, i.e. the relative deformation of the construct after which it did not return to its original, pre-loaded form, was calculated as the difference between displacements observed between the initial and final cycles of loading. The final elastic deflection of each screw was defined as the displacement observed within the final cycles of loading, essentially representing a construct that had been deformed by prior loading cycles, but was now consistently returning to its original form. The total displacement of each construct was defined as the difference in displacement from neutral, i.e. 0, and the last cycle (Fig. 3). Each construct was again axially loaded to failure after cyclical loading.

Fig. 3.

Hysteresis curve demonstrating the first, and final ten cycles of axial loading from partially threaded screw number 2. The plastic deformation, final elastic deflection, and total displacement are calculated as above. The graph is representative of an axial load being applied, causing an initial deformation. The construct is then unloaded, resulting in an elastic return to its original shape. The plastic deformation can be interpreted as permanent construct deformation, i.e. the construct is no longer returning back to its original state as it has changed morphology. Of note, the final 10 cycles can be seen to be overlapping each other directly, demonstrating that the construct has essentially “settled”, representing elastic changes during cyclical loading.

Student’s 2 tailed t-test was performed in MATLAB® using the “ttest” function to detect differences between the properties of the two types of screw, at a significance level of 0.05.

3. Results

3.1. Maximal insertional torque

Utilising the MTS, the fully threaded screw achieved an MIT of 11.22N-m, compared to 2.15N-m for the partially threaded screw, which in turn returned insertional torque values of 7.85N-m for fully threaded screws, and 1.50N-m for partially threaded screws (Fig. 2). Again, we were unable to manually reproduce an insertional torque of 7.85N-m for the fully threaded screws and therefore 6.58N-m was used.

Fig. 2.

Graph demonstrating maximum torque readings obtained with screw rotation. The step-offs seen on each graph line were as a result of the MTS only being able to turn 270°, requiring the system to be reset and thus dropping the observed torque to 0.

3.2. Pre-cyclical loading

In axial compression, the fully threaded construct had significantly higher yield force (1727.60 N vs 326.96 N, p = 0.0092), stiffness (761.55 N/mm vs 473.77 N/mm, p = 0.0061), ultimate force (3489.92 N vs 1549.53 N, p = 0.019), and yield deformation (2.50 mm vs 0.86 mm, p = 0.025) when compared to partially threaded constructs (Fig. 4).

Fig. 4.

Histograms with standard error bars demonstrating ultimate force (N), stiffness (N/mm), yield force (N), and yield deflection (mm), for partially and fully threaded screws before and after cyclical loading.

3.3. Post-cyclical loading

During cyclical loading of the fully threaded construct, the average plastic deformation was 0.041 mm ± 0.005. The average plastic deformation observed for partially threaded construct was 336% higher at 0.136 mm ± 0.087; however, this difference did not reach statistical significance (p = 0.070). The final elastic deflection between the two types of construct was not statistically significant (0.84 mm vs 1.013 mm, p = 0.25). The mean total movement of partially threaded constructs was 21% higher (1.013 mm, ±0.16 vs 0.84 mm, ±0.097, p = 0.25). (Fig. 5).

Fig. 5.

Histograms with standard error bars demonstrating plastic deformation, final elastic deformation, and total movement, for partially and fully threaded screws following cyclical loading.

Following cyclical loading, fully threaded constructs had a significantly higher average yield force (1923.05 N vs 608.21 N, p = 0.00011), ultimate force (4038.20 N vs 1658.48 N, p = 0.00037), and yield deformation (2.65 mm vs 1.07 mm, p = 0.0016) (Table 1). The stiffness of fully threaded constructs and partially threaded constructs was not significantly different following cyclical loading (802.57 N/mm vs 787.49 N/mm, p = 0.3). When comparing constructs that had or had not been cyclically loaded, the stiffness and yield force of the partially threaded screws increased significantly.

Table 1.

Summary of results.

	Partially Threaded	Fully Threaded		p-value
Pre-Cyclic Axial Loading
Ultimate Force	1.55 ± 0.14	3.49 ± 0.45	kN	0.019
Stiffness	0.47 ± 0.04	0.76 ± 0.00	N/mm	0.006
Yield Force	0.33 ± 0.17	1.73 ± 0.15	kN	0.009
Yield Deflection	1.86 ± 0.47	2.50 ± 0.16	mm	0.025
Cyclical Loading
Total Movement	1.01 ± 0.16	0.84 ± 0.10	mm	0.107
Plastic Deformation	0.14 ± 0.09	0.04 ± 0.01	mm	0.071
Elastic Deflection	0.85 ± 0.06	0.77 ± 0.09	mm	0.248
Post-cyclical Axial Loading
Ultimate Force	1.66 ± 0.22	4.04 ± 0.18	kN	0.000
Stiffness	0.79 ± 0.01	0.80 ± 0.02	N/mm	0.300
Yield Force	0.61 ± 0.03	1.92 ± 0.09	kN	0.000
Yield Deflection	1.07 ± 0.05	2.65 ± 0.25	mm	0.002

4. Discussion

It is accepted that early operative treatment of a displaced posterior pelvic ring allows for early rehabilitation and reduces morbidity.¹² The ideal technique and construct for sacro-iliac or transiliac-transsacral screw fixation remains unknown due to the variety of posterior pelvic fracture patterns and significant forces transmitted through the SI joints.^7,13,14 The current literature regarding fully and partially threaded SI screws investigates the incidence of iatrogenic nerve injury. There is no indication that a compression screw puts the neurovascular structures at risk.^15,16 However, late displacement or loss of fixation can have significant functional consequences.^1,4,17 No studies to date have compared the biomechanical properties of a fully threaded lag-by-technique posterior pelvis screw against a lag-by-design partially threaded screw.

Screw loosening typically occurs early in the rehabilitation period following treatment of unstable pelvic ring injuries, as early mobilisation after surgery is encouraged if no other injuries preclude weight bearing. Kim et al., in 2016 found a 17.3% incidence of sacroiliac screw loosening at a mean of 25.3 days (range, 10–70 days), and revision rate of 11.8% in their retrospective review of 110 pelvic ring injuries that received percutaneous fixation with a 7.0 mm partially threaded design.⁷ The incidence of loosening was found to be significantly higher in vertical shear (VS) type fractures (29.7%), when compared to anterior-posterior compression (11.1%), or lateral-compression (10.9%) type fractures (p = 0.014). Screw fixation within the middle 1/3 of the S1 body had a higher screw loosening when compared to screws abutting the anterior cortex of the S1 body (23.4% vs 8.7%, p = 0.044). Patients with VS fractures and concomitant zone-2 sacral fractures were also significantly more likely to experience screw loosening when compared to those sustaining VS fractures with zone-1 sacral fractures, or simple sacroiliac joint dislocation (43.5% vs 7.1%, p = 0.019). These findings were similar to those observed by Griffin et al. in their retrospective review of 62 patients sustaining VS type fractures, treated with a partially threaded iliosacral screw, who found that a vertical sacral fracture pattern permeated an excessive risk of failure of 13%.¹⁴ Within their cohort, all of the failures occurred within 3 weeks of the initial surgery. The morphology of the injury as well as surrounding bone stock also contributes towards construct failure, thus underscoring the need to maximize fixation strength and stability.

Anecdotally, the senior author (MM) has noted subjective increase in screw purchase and insertional torque when placing fully threaded screws in vivo. We found a higher maximal insertion torque prior to failure that was more than five times greater for the fully threaded lag screw when compared to the partially threaded screw. The two types of stainless steel screws compared in this study have the identical thread and core diameters, 7.3 mm and 5.0 mm respectively, with a thread pitch of 2.75 mm. The torsional strength of a screw, i.e. how much torque it can withstand prior to plastic deformation, is proportional to the cube of the core diameter as well as the material from which it is made.¹⁸

Interestingly, the insertional torque of a screw has not been correlated to the pullout strength. Ricci et al. investigated the effect of insertional torque on the pullout strength of cortical screws with varying pitches (range 1.0 mm–1.75 m) within a cancellous bone surrogate. The MIT obtained for larger pitch screws was significantly higher for those obtained with screws with a pitch of 1.2 mm or less, however, their findings concluded that screw pitch, as well as the screw’s ability to generate a higher initial insertional torque were not predictive of pullout strength.¹⁹ Tankard et al. in a cadaveric study of 15 pairs of humeri stratified to normal, osteopenic, or osteoporotic groups, found that the pullout strength for both normal and osteoporotic bone was greatest at 50% of the MIT, and not statistically significant from pullout strengths at 70–90% of the MIT. For osteopenic bone, the pullout strength was greatest at 70% of MIT but the absolute value itself was not significantly different from the pullout strengths at 50% and 90% of MIT.²⁰ The aforementioned studies did not introduce any micromotion within the construct prior to testing the pullout strength, which has been shown to inhibit the formation of bone, instead becoming enveloped in a fibrous tissue which is also evident radiographically by a visible “halo”.²¹

The utility of the screws used for sacroiliac fixation are to prevent axial displacement and rotation.¹³ The results of our axial loading tests are similar to those seen by Downey et al. with the fully threaded screws resisting much higher deforming forces prior to failure when compared to the partially threaded screws.⁶ The authors found no difference between the stiffness of the two screws. Our findings demonstrate that prior to any loading, the partially threaded screw exhibited a significantly lower stiffness, which increased following cyclical loading. We propose that this phenomenon is due to the bone surrogate essentially settling around the screw, as the stiffness of the two screws following cyclical loading was strikingly similar, which is to be expected given the identical material properties of the screws. Despite the differences in plastic deformation seen between the two constructs, there was no difference in the total displacement, which we believe is due to the aforementioned reason, as the hysteresis curves for the final cycles are visibly identical.

Prior studies investigating screw loosening under dynamic loading conditions have utilised plate constructs, which is not applicable in the management of sacroiliac disruption.22, 23, 24 The length of thread engagement, as well as the diameter of the screw have also been correlated with the holding strength of cancellous screws in biomechanical models, as these contribute directly to the thread-bone interface, through which failure occurs by shear.^25,26 Given that the two screws are manufactured from the same material and the only differing factor is the length of threads, we propose that the increased length of thread engagement led to the difference in forces seen, and that the initial micromotion upon initial loading may contribute to ultimate screw failure in vivo.

There were several limitations in the present study, consistent with biomechanical modelling. Firstly, our construct is not representative of the sacro-iliac joint, given the surrounding ligamentous contribution towards stability, and thus the results of this study cannot truly be applied clinically. Furthermore, we recognise that posterior fixation of pelvic ring injuries is often performed with anterior stabilisation, however the constructs used were not intended to be representative of a physiological pelvic model, but rather to identify specific screw characteristics in vectors of loading. The torque results in our study were obtained with the utilisation of a manufactured washer, as we found that during screw insertion the head and washer would subside into the testing material. Furthermore, the MIT obtained for the fully threaded screw with use of this washer far exceeded the torque able to be generated manually. For this reason, statistical testing was not performed for torque values as the results obtained cannot be extrapolated into clinical practice. However, the same washer was utilised for all constructs, and although torques obtained may not be representative of that obtained in vivo, they are still consistent and reliable for the purpose of this study. The orientation of the screw insertion may have differed between constructs, which may explain the large standard deviation seen in the plastic deformation of the screws. This phenomenon could be reduced with an infinitely larger sample size, far beyond what would be possible in vivo. Furthermore, our testing setup did not allow us to identify whether or not the displacement was occurring at the implant or the compression of the testing material itself; however, our intention for this study was not to obtain absolute yield force and displacement characteristics, but rather provide a quantified comparison of loading characteristics between the two types of screw within a consistent medium.

5. Conclusion

The present biomechanical study supported the hypothesis that fully threaded lag screws were able to withstand significantly higher axial displacement loads and insertional torque when compared to a partially threaded screw in an SI joint model. In addition, fully threaded lag screws exhibited a trend toward smaller plastic deformation, i.e. irreversible deformation, when compared to partially threaded screws, although this did not reach significance. Given the extreme forces placed across the sacroiliac joint with weight bearing, the ideal fixation construct minimizes displacement, This biomechanical model indicates that the increased thread engagement length in fully threaded lag screws within isolated cancellous bone may be an advantage over more commonly used partially threaded screws. These promising results underscore the need for well modelled cadaveric studies and subsequent clinical trials to further elucidate the ideal screw construct in unstable pelvic ring injuries.

Ethical approval

IRB approval/Research Ethics Committee approval was not required for this original article.

Funding

There were no sources of funding for this original article.

The authors have received internal grant support from the Emory University School of Medicine.

The authors do not have any proprietary interests in the materials described in the article.

Declaration of competing interest

One or more of the authors report personal fees outside of the submitted work.