Pullout strength of pedicle screws with cement augmentation in severe osteoporosis: A comparative study between cannulated screws with cement injection and solid screws with cement pre-filling

Osteoporosis is a common disease in aging populations. Spinal surgeons unavoidably encounter patients with osteoporosis who need spinal decompression and instrumentation due to degenerative spinal diseases. However, pedicle screw instrumentation in a severely osteoporotic spine remains a challenge for orthopedic surgeons.

Previous studies have demonstrated that the holding power of screw in non-augmented osteoporotic bone decreases with decreasing bone mineral density [1‐3]. Consequently, to date, efforts to improve screw holding power have focused primarily on the pullout of screws augmented with polymethylmethacrylate (PMMA) [4‐8], calcium phosphate [9‐11] and calcium sulfate [12, 13] in osteoporotic bone.

PMMA augmentation is regarded as an efficient method to enhance screw strength in osteoporotic bones [4‐8]. Traditionally, to improve the anchoring strength of screws in osteoporotic bone, PMMA is injected directly into the prepared pilot hole of the vertebral body prior to screw insertion. The pedicle screw is then inserted into the cement to enhance the screw anchoring strength. Another insertion technique involves the usage of an expandable screw, which allows for flange expansion at the screw tip and, hence, increases the screw holding power [14, 15]. Recently, works have focused on the perforated screw with PMMA augmentation, which allows for the injection of cement through the perforation to achieve the improvement of screw anchoring strength [4, 16, 17]. Although numerous studies address the improvement in pullout strength with various screw augmentation techniques, a comparison of screw insertion technique between solid screws with retrograde cement pre-filling and cannulated screws with cement injection through perforation is lacking.

Our review of the literature found that most of the current research on pedicle screw pullout has been performed with cylindrically shaped pedicle screws [4, 5, 7, 18‐20]. Cross sections of pedicles at the lumbar spine have an elliptical shape. In the anterior-posterior direction, the pedicle margin converges [21]. Conical screws were developed to better match the anatomical situation in the pedicle. Conically shaped screws were demonstrated to provide better fixation strength when compared to cylindrically shaped screws [22, 23]. Theoretically, conical screws should progressively compress the surrounding bone with each turn of the screw during insertion. This compression of the surrounding bone should be beneficial, as it provides increased purchase during screw insertion [24, 25]. However, as a result of the rapid reduction in the compression of the surrounding bone if the screw is partially removed to adjust the screw placement during surgery, the reduction of the compression stress could also cause a rapid reduction in the fixation strength. The effect of partial screw removal on the pullout strength is a concern for surgeons because it is often necessary to adjust the insertion depth during screw placement.

Studies have investigated the mechanical performance of pedicle screws in vertebrae, but only minimal data are available concerning the performance of conical screws in the presence of compromising events such as partial screw removal. In the present study, we address two features of the pullout strength of conical versus cylindrical screws: 1) the fixation strength between solid screws with retrograde cement pre-filling and cannulated screws with cement injection through perforation and 2) the effect of partial screw removal after full insertion.

The radiological and physical examinations of the screws inserted into the test blocks and specimens after the pullout tests are shown in Figure 3. The radiological photographs (Figure 3, top) indicated that the area of the cement/screw interface was greater for solid screws with retrograde cement pre-filling than for cannulated screws with cement injection. Observations of the failed specimens after the pullout test (Figure 3, bottom) indicated that cement infiltration into the open cell of the test block led to formation of a (cement/bone) composite structure. All of the failures occurred at the composite/bone interface; however, the composite remained well bonded to the screws, implying that the screw/composite interfacial strength was much higher than the composite/bone interfacial strength.

The average ultimate pullout strengths of fully inserted conical and cylindrical pedicle screws for the different cement augmentation techniques are shown in Table 3 and Figure 4. Regardless of the screw outer geometry (conical or cylindrical), solid screws with retrograde cement pre-filling exhibited significantly higher pullout strength than that of the cannulated screw with retrograde cement pre-filling (p = 0.0129 for conical screws; p = 0.005 for cylindrical screws); whereas solid screws without cement augmentation exhibited the lowest pullout strength (p < 0.001). For the conical screws, the solid type provided a 23% increase in the pullout strength when compared to the cannulated type (p = 0.0129). In contrast, for the cylindrical screws, the solid type provided a 41% increase in the pullout strength when compared to the cannulated type (p = 0.005). For a given screw augmentation technique (screw without cement augmentation, cannulated screw with cement injection or solid screw with cement pre-filling), no significant difference in pullout strength was found between conical and cylindrical screws (p > 0.05).

The average ultimate pullout strength for conical and cylindrical pedicle screws tested at full insertion and after a 360-degree screw backout is shown in Figure 5. Regardless of the screw outer geometry (conical or cylindrical), the pullout strengths were unchanged (not significant) after the partial removal from full insertion. Additionally, the pullout strength for pedicle screws after partial removal had a larger standard deviation than that after full insertion.

The Taguchi analysis indicated that the design factor "cement augmentation technique" was the main influential factor, whereas "screw shape" was the least influential factor affecting the pullout strength.

Adequate pedicle screw fixation in the presence of compromised bone quality presents a challenging problem for spine surgeons. Cancellous bone is more profoundly affected by osteoporosis process than cortical bone, and it is not surprising that the anchoring strength is significantly decreased in patients with low bone mineral density.

Cancellous bone is generally reported to have a density in the range of 0.09 to 1.25 g/cm³[26]. The variations in apparent density, trabeculae orientation and mechanical properties of cancellous bone within and between specimens are large. Consequently, a very large number of tests are required to isolate the effects of screw design. The use of synthetic cancellous bone simplifies the experimental set-up, thus limiting experimental error. In the present study, commercially available artificial osteoporotic bones with an open celled structure (test blocks) were used as a substitute for human osteoporotic vertebrae. The manufacturer's instruction states the following: The appearance of open cell rigid foam resembles that of human cancellous bone [28]. The test blocks offer a uniform and consistent density that eliminates the variability encountered when testing with human cadaver bones [31]. Consequently, test blocks are suitable for a variety of applications requiring an open-cell structure, such as cement injection and the modeling of osteoporotic cancellous bone. Many biomechanical studies using polyurethane foam (test block) to simulate osteoporotic cancellous bone have shown that polyurethane foam is a good alternative for in vitro testing [3, 27, 29, 32]. In recent studies, Hashemi et al. [32] compared the axial pullout strength and insertion torque of augmented and nonaugmented pedicle screws using polyurethane foam with densities of 0.16 and 0.32 g/cm³, corresponding to the porosity of osteoporotic and normal bones, respectively. Their results revealed a significant correlation between peak pullout resistance and insertion torque. Zehnder et al. [29] investigated the effects of screw orientation and load to failure of a plate/bone construct that was attached to 0.09 g/cm³ polyurethane foam (used to simulate severely osteoporotic cancellous bone). They concluded that in a severely osteoporotic model, failure in cantilever bending at low forces would take place regardless of fixation methods used and the added benefit of oblique screw placement observed in healthy bone is not observed in osteoporotic bone. In addition, Ramaswamy et al. [3] compared the pullout strengths of four different commercially available cannulated screws inserted in polyurethane foam blocks with three different densities (0.16, 0.24 and 0.32 g/cm³) simulating osteoporotic, osteopenic and normal bones, respectively. Their results indicated that the holding power of screws is directly correlated to bone density, thread design and number of threads engaging the bone. Reinsertion through the same hole could reduce the ultimate pullout strength. Additionally, Patel et al. [27] performed the compressive test on different densities (0.32, 0.16 and 0.09 g/cm³) of polyurethane foams to examine whether the commercially polyurethane foams are suitable for mimicking human osteoporotic cancellous bone. The fracture stresses of these foams enable them to be used as models for normal (0.32 g/cm³), osteoporotic (0.16 g/cm³) and very low density osteoporotic cancellous bone (0.09 g/cm³). They concluded that the 0.16 g/cm³ polyurethane foam is a good alternative for in vitro testing because it has compressive Young's modulus and yield strength values similar to osteoporotic bone that has also been tested in compression. In the current study, in considering of the ease of cement injection, the polyurethane foam with a density of 0.09 g/cm³ was chosen for emphasis of extremely osteoporotic cancellous bone. We believe that although the pullout test did not measure the actual screw/bone interfacial strength, the 0.09 g/cm³ polyurethane foam provides a uniform platform to compare the mechanical behavior of pedicle screws with various designs.

Few reports have addressed the effects of partial screw removal on the bone/screw interfacial strength [22, 33]. However, the reported results have been inconsistent. Abshire et al. [22] investigated the characteristics of pullout failure in conical and cylindrical pedicle screws after full insertion and partial removal using porcine lumbar vertebrae. They concluded that there was no reduction in the pullout strength, stiffness or work to failure when pedicle screws were partially removed either by 180 or 360 degrees, from full insertion. Lill et al. [33] examined the mechanical performance of cylindrical and dual-core pedicle screws that were fully inserted and then removed 4 mm from calf and human vertebrae. Their results indicated that partially removing the screws before cyclic loading led to an increase in displacement of 32%, which resulted in a significant reduction in the screw pullout strength. In the current study, we report that back-out of the screw by 360 degrees did not reduce the pullout strength; our results were inconsistent with those of Lill et al. Reasons causing the insistent results might be attributed to the fact that in their investigation into the effect of screw removal in calf and human vertebrae, PMMA cement was not used and severe osteoporosis was not considered. Another factor causing the inconsistent results might be the absence of the cortex shell in the current study. Nevertheless, our results indicate that in a synthetic material with a density similar to that of severely osteoporotic bone, screw back-out does not affect the pullout strength of screws with cement augmentation.

PMMA cement infiltration into the open cell of the test block led to the formation of a composite cement/bone structure in the area of cement infiltration (Figure 3). For all of the cases, observation of the failed specimens revealed that the failure occurred at the composite/bone interface, whereas the composite remained well bonded to the screws (Figure 3, bottom). In the present study, for a given screw design (conical or cylindrical), we report a significantly higher pullout strength for solid screws with retrograde cement pre-filling than for cannulated screws with cement injection using artificial osteoporotic bones (Figure 4). Our results demonstrate that solid screws with retrograde cement pre-filling offer improved initial screw fixation strength in severe spinal osteoporosis. The radiological examinations indicated that the area of the composite/bone interface was greater for the solid screws with retrograde cement pre-filling than for the cannulated screws with cement injection (Figure 3). This increased interface led to significantly higher pullout strength for solid screws. For the cannulated screws, the distal placement of the radial holes provided a longer distance for the cement to flow outside from the distal end of screw. This would result in most cement being distributed on the distal end of the screw and would reduce the area of the composite/bone interface enormously. In contrast to the cannulated screw, cement is prefilled prior to screw insertion for solid screws. During screw insertion, the prefilled cement is squeezed to occupy some of the voids of the adjacent trabecular bone, which distributes cement on the more proximal threads of the solid screw because of a host of factors such as cement wettability, porosity considerations, flow rate, etc. This causes a difference in density between the bone infiltrated with cement and the adjacent intact trabecular bone. The enormous difference in density between composite structure and intact bone was thought to induce a general failure mode along the composite/bone interface during axial screw pullout. Theoretically, the pullout force required to remove the composite structure (bone with cement infiltration) from adjacent intact trabecular bone is proportional to the composite/bone contact area. Consequently, we believe that a greater area of the composite/bone interface will be beneficial to the anchoring strength of the screw. Although cement pre-filling prior to screw insertion benefits the pullout strength, cement leakage from the screw insertion point is more likely with cement pre-filling (Figure 3). This result implies that the use of solid screws with retrograde cement pre-filling technique has an associated risk of cement leakage into the spinal canal.

The Taguchi method utilizing an orthogonal array has proven to be a useful tool to markedly reduce the total number of required experiments. The method contains a well-chosen subset of all possible test condition combinations and can achieve a balanced comparison of levels of any factor [34‐36]. In the current study, although all factorial trials had been performed, a further investigation using the Taguchi method to determine the main contribution of the design factors was appropriate. The Taguchi results indicated that, rather than eight trials, only four trials were required to precisely determine the most influential factors. The Taguchi's L₈ array analysis revealed the design factor "cement augmentation technique" was the main influential factor, whereas "screw shape" was the least influential factor affecting the pullout strength. In addition, the highest pullout strength was found in combination of cylindrical-solid-full insertion. The results of Taguchi analysis were consistent with our conclusion that the PMMA augmentation technique for solid screws with retrograde cement pre-filling offers improved initial fixation strength.

Our study is an in vitro analysis of specimens prepared in a laboratory environment, which does not necessarily represent clinical circumstances. There are limitations to this study. First, a test block was used as a substitute for human osteoporotic vertebrae. Although the synthetic bone provides a platform for comparison of pullout strength, the material properties of the test block are somewhat different from those of actual osteoporotic cancellous bone; the extrapolation of our results to clinical utilization should be performed with caution. Second, the measurement of pullout strength at the screw-bone interface did not take into account the cortical shell of the spinal vertebrae, which may have an impact on interfacial bonding strength. However, our specimens were all prepared and tested in a uniform, reproducible manner, and we believe that our results provide a comparison of the mechanical performance of various screws in severely osteoporotic bone. Third, the volume of injected cement tested was constant (3 ml). The amount of injected cement might be an important influential factor in determining the screw holding power. The effects of the amount of injected cement on bone/screw interfacial strength deserved to be conducted in the future. Last, the present work is limited to static loading (pullout in test block) without considering of other physiological loadings. In actual physiological situations, the screw/bone interface is subjected to complex dynamic multi-directional loading. Our results cannot be used to predict the biomechanical performance of screw fixation under cyclic loading in the long term. Therefore, the possible future work could be to investigate the fatigue properties of pullout strength of the solid vs. cannulated pedicle screws in animal models.

We conclude that solid screws with retrograde cement pre-filling offer improved initial fixation strength when compared to that of cannulated screws with cement injection through perforation for both conically and cylindrically shape screws. Our results also suggest that the fixation screws can be backed out by 360 degrees for intra-operative adjustment without the loss of fixation strength.