Traditional lumbar pedicle screw fixation has become a common surgical procedure for the treatment of lumbar degenerative diseases, and one of the most important factors affecting pedicle screw strength (pull-out resistance) is screw placement trajectory [7]. Most of the screw path are located in the vertebral cancellous bone. Screw loosening is highly common as BMD decreases, which weakens the fixation effect [8]. Therefore, it is necessary to determine the bone density of the screw placement area before the operation in order to select the screw placement track with better bone condition [20]. At present, the position of the insertion point of the PLIF is acknowledged, located at the junction of the midline of the transverse process in the posterior midsection of the vertebral body and the outside of the facet joint [13]. This allows the trajectory of the screw at the pedicle and the posterior aspect of the vertebral body almost the same after being inserted into the vertebra. Although the direction of screw placement for most surgeries is parallel to the endplate, due to the influence of many factors, such as the curvature of the patient’s lumbar spine, the model of the implanted fusion cage, etc., the trajectory direction of the screw head at the anterior aspect of the vertebral body has a certain degree of variability [21]. At the same time, it is not clear whether screw trajectories in different directions can obtain the same pull-out resistance. We think that it is necessary to analyze the HU values (indirect BMD results) in the anterior region of the screw inserted at the anterior aspect of the vertebral body. Therefore, this study is the first to measure the HU values around the screw path of the lumbar pedicle screws in different directions at the anterior aspect of the vertebral body and analyzes the optimal screw placement direction to obtain the optimal pull-out resistance.
According to the overall results of all patients in this study, the parts with higher HU values were located in the middle 1/3 of L1 and L2, the lower 1/3 of L4, and the changing trend of HU values in different parts of L1-L5 vertebral body was higher in the middle 1/3 of upper lumbar vertebra and higher in the lower 1/3 of lower lumbar vertebra. The HU values of different segments from L1 to L5 were the lowest in L3 and the highest in L5, which was consistent with the results of previous studies on the mean HU values of each vertebral body [14]. It was considered that the middle 1/3 of L1 was the most favorable for screw placement, with the direction parallel to the endplate. It was also found in the study that the younger the patients or the better the BMD, the more significant the difference of HU values in different parts of the vertebral body was. The HU values in the middle 1/3 of L1 and L2 was higher, while the HU values in the lower 1/3 of L3-L5 was higher. Combined with these results, we hypothesized that the transition of high HU from the middle of the upper lumbar spine down to the lower lumbar spine might be due to physiological and biomechanical reasons. These young patients with better bones did not have more serious structural changes in the lumbar spine. Therefore, their high HU values of L1 and L2 were concentrated in the middle of the vertebral body to maintain the stability of the entire lumbar spine in a more balanced way, while the lower lumbar spine, especially the lower part of the lower lumbar spine, carried a large portion of the body weight to support normal and weight-bearing activities, so the HU value was relatively increased [22]. At the same time, we observed a deterioration in bone mass and a decrease in the mean HU value of each vertebral body with increasing age, which was consistent with the findings of Schreiber JJ [9]. In addition, it was also found in our study that the differences of HU values in different parts of the vertebral body also decreased, and this change first appeared in the lower lumbar (L4, L5), and involving the upper lumbar gradually, finally only osteoporosis group without the high HU value in the middle of L1 results, showing the whole lumbar parts HU values were no difference. We considered that such a result might be due to the fact that the older the person was, the more serious the degeneration of bone was, and the more likely and serious the degenerative changes of the structure of the lumbar spine were to occur, while L4 and L5 were the more prone and easily affected first [23]. As a result, the degeneration of the lower lumbar spine resulting in the decrease of the HU value difference in various parts of the vertebral body was earlier and more obvious, so that the degeneration of the entire vertebral body to a low HU value was the first manifestation. From the perspective of different genders and different diseases, the results were almost consistent with the overall performance. The pronounced reasons for the subtle differences remain unclear and require further investigation. We analyzed that this difference might be related to other factors, such as different lifestyle, physical activity, and different changes in the vertebral body or endplate. The analysis reason might be that the difference was not obvious due to the small number of cases.
Previous related studies have suggested that the value of BMD around the screw track can affect or even predict the loosening of the screw [24]. DEXA is the current gold standard for BMD measurement and osteoporosis screening. However, the accuracy of BMD measurement by DEXA is affected by degenerative changes of the lumbar spine, scoliosis, osteogeny, compression fracture, calcification and other aspects, and cannot distinguish between cortical bone and cancellous bone. Some studies have found that the detection of BMD based on the HU value measured not only had a certain linear relationship with the BMD evaluated by DEXA and QCT, but also could be used for auxiliary diagnosis of osteoporosis, while avoiding the above shortcomings [9]. It is noteworthy that many new reports about the clinical application of HU were emerged in recent years. Matsukawa et.al [11] found that the pull-out resistance of the screw was correlated with the HU by measuring the screw path HU and the insertion torque of the inserted screw. With a low HU value, the pull-out resistance of the screw is also low, indicating the risk of screw loosening. Many other similar studies have also concluded that preoperative CT measurements of HU could predict screw loosening [25, 26]. Zhang.et.al [27] compared the HU values of the entire screw trajectory between the cortical screw and the traditional screw, concluded that the pull-out resistance of the cortical screw was stronger. Wichmann et.al [28] found that the bone mineral density at the pedicle was more strongly correlated with the pull-out resistance of the inserted screws compared with that inside of the vertebral body. Meanwhile, it was also found that the bone mineral density of the pedicle screws in the vertebral body segment (the area of the screw head) was significantly different from that at the pedicle, which was lower than that at the vertebral pedicle. On this basis, Fei Xu et.al [29] respectively analyzed the relationship between the HU value at the pedicle and inside the vertebral body and screw loosening, and also found that the HU value at the pedicle was more correlated with screw loosening than the HU value in the vertebral body, and the prediction was more sensitive. However, it was also found that the HU value in the vertebral body had the same predictive value. The results of the research of these scholars have confirmed that the value of HU (around the vertebral body or screw path) related to the loosening and pull-out resistance of the screw. But to date, no study has been conducted on the basis of these findings to select a better and more suitable screw placement direction. Therefore, we believe that the analysis of the screw trajectory in different directions, especially the screw head area in the anterior part of the vertebral body, may provide new findings, which can provide some reference for the selection of a suitable screw placement direction.
This study has the following deficiencies. Firstly, this study is a retrospective study, and more prospective studies are needed to verify the relationship between HU values in different directions of the vertebral body and pull-out resistance after screw placement. At the same time, these patients need to be further followed up clinical and radiological follow-up data to assess the screw placement status. Secondly, we measured the HU value by manually selecting ROI. Although in this study, multiple measurements and averaging were used to improve inter-observer and intra-observer reliability, this might cause concerns about the repeatability of these results. In addition, we did not conduct a formal BMD measurement for the patients, but relied on the HU conversion formula to convert into QCT values, rather than the bone density values measured by DEXA as the diagnostic criteria, so as to determine patients with different bone conditions, which might lead to diagnostic errors. Finally, we measured the HU values in three of the cross-sections after trisection of the vertebral body, which represented regional HU values, reflected the local two-dimensional bone mineral density, in the actual operation of screw placement is three dimensional, so by only a few cross-sectional data might obscure or ignore other potential factors that affect the HU.
Based on the above analysis, we can consider to improve the direction of screw placement inside the vertebral body preoperatively. According to our statistical results, the sagittal placement direction of pedicle screws in patients with L1 and L2 lesions can be parallel to the endplate, and the patients with L3, L4 and L5 lesions can be appropriately caudally tilted (Fig. 7). Additionally, we conducted a separate stratified analysis for patients with lumbar spondylolisthesis and osteoporosis. Osteoporosis is one of the important risk factors for osteoporotic vertebral compression fractures [30], and VCF (vertebral compression fracture) is also a common postoperative complication of PLIF in spine surgery [31]. Besides, previous studies have confirmed that lumbar spondylolisthesis is an independent risk factor for osteoporotic vertebral compression fractures, and lumbar spondylolisthesis is not only associated with disc degeneration but also with curvature of the spine, such as lumbar lordosis or chest kyphosis increase, which could affect sagittal curvature or spinal sagittal balance and cause VCF [32]. These patients need to use internal fixation surgery for reduction to improve symptoms and function, and obtain stability, so a higher requirement for screw pull-out resistance is often needed to further avoid surgical failure. Although there was almost no difference in different parts of the vertebral body for elderly patients, osteoporosis, and patients with lumbar spondylolisthesis and osteoporosis (Fig. 6c), the results were found to be almost identical, which may also be related to the small number of samples by analyzing the variation trend of HU values in these patients and the overall sample, and the standard we concluded could still be used for evaluation. Therefore, cortical bone screws, injectable fillers at the surgical site or adjacent vertebral segments and adjusting the size and shape of screws should be considered for aged patients and osteoporotic patients according to the actual situation of the operation to ensure a better pull-out resistance of the screws. Simultaneously, special attention should be paid to the patients with lumbar spondylolisthesis and osteoporosis to prevent vertebral fracture and other postoperative complications.