Establishment and Verification of a Predictive Nomogram for New Vertebral Compression Fracture Occurring after Bone Cement Injection in Middle‐Aged and Elderly Patients with Vertebral Compression Fracture

Objective New vertebral compression fracture (NVCF) occurring after bone cement injection in middle‐aged and elderly patients with vertebral compression fracture is very common. Preoperative baseline characteristics and surgical treatment parameters have been widely studied as a risk factor, but the importance of the patients' laboratory indicators has not been thoroughly explored. We aimed to explore the relationship between laboratory indicators and NVCF, and attempt to construct a clinical prediction model of NVCF together with other risk factors. Methods Retrospective analysis was performed for 200 patients who underwent bone cement injection (percutaneous kyphoplasty or vertebroplasty) for vertebral compression fractures between January 2019 and January 2020. We consulted the relevant literature and collated the factors affecting the occurrence of NVCF. Feature selection of patients with NVCF was optimized using the least absolute shrinkage and selection operator regression model, which was used to conduct multivariable logistic regression analysis, to create a predictive model incorporating the selected features. The discrimination, calibration, and clinical feasibility of the predictive model were assessed using the concordance index (C‐index), calibration plots, and decision curve analysis. Internal validation was performed using Bootstrap resampling verification. Results Time from injury to surgery exceeding 7 days, low osteocalcin levels, elevated homocysteine levels, osteoporosis, mode of operation (percutaneous vertebroplasty), lack of postoperative anti‐osteoporosis treatment, and poor diffusion of bone cement were independent risk factors for NVCF in middle‐aged and elderly patients with vertebral compression fracture after bone cement injection. The C‐index of the nomogram constructed using these seven factors was 0.895, indicating good discriminatory ability. The calibration plot showed that the model was well calibrated. Bootstrap resampling verification yielded a significant C‐index of 0.866. Decision curve analysis demonstrated that the greatest clinical net benefit for predicting NVCF after bone cement injection could be achieved with a threshold of 1%–91%. Conclusion This nomogram can effectively predict NVCF incidence after bone cement injection in middle‐aged and elderly patients with vertebral compression fracture, thus aiding clinical decision‐making and postoperative management, promoting effective postoperative rehabilitation, and improving the quality of life.


Introduction
V ertebral compression fractures caused by minor trauma or no obvious trigger are the most common fragility fractures. 1 Pain and limitation of movement caused by fracture are the principal factors affecting the quality of life, and may even result in disability in middle-aged and elderly patients. 2 Due to the aging of the population, vertebral compression fracture has led to myriad medical and healthcare problems. 1 Currently, percutaneous kyphoplasty (PKP) and percutaneous vertebroplasty (PVP) are the most frequently used and successful minimally invasive surgical procedures for vertebral compression fractures, which can effectively ameliorate pain and facilitate rapid recovery. [3][4][5] However, case reports and series describing postoperative new vertebral compression fracture (NVCF) have garnered considerable attention, owing to the popularity of PKP and PVP. According to studies, the prevalence of NVCF is approximately 2%-23% after PKP and 2.4%-52% after PVP. 6 Meanwhile, NVCF, as one of the most serious complications after bone cement injection, may lead to more severe local pain, which brings a huge psychological and economic burden to patients. Therefore, more clinicians and scholars have begun to pay attention to it and explore the occurrence and development of NVCF. 6 However, the cause of NVCF remains unclear. The known risk factors for NVCF include preoperative demographic characteristics such as osteoporosis and advanced patient age, and surgical factors such as bone cement leakage and lack of anti-osteoporosis treatment. 7,8 Nevertheless, laboratory indicators such as osteocalcin and homocysteine that can reflect the condition of bone metabolism have not been studied in detail.
Therefore, this study aimed to investigate laboratory indicators, preoperative baseline characteristics, and surgical treatment-related parameters as possible risk factors, to explore the relationship between them and the occurrence of NVCF, and to construct a predictive nomogram. A nomogram is a predictive model that estimates the risk of a specific disease by integrating several factors for complex operations. 9 It can aid in comprehensive evaluation of the probability of postoperative NVCF, and implementation of targeted and personalized preoperative and postoperative management for high-risk patients, with the objective of reducing the incidence of postoperative NVCF.

Patient Population
This study entailed a retrospective analysis of 200 patients who underwent bone cement injection (PKP or PVP) for vertebral compression fractures at the Affiliated Hospital of Shandong University of Traditional Chinese Medicine between January 2019 and January 2020. All patients were monitored for 2 years. The study protocol was approved by the Affiliated Hospital of Shandong University of Traditional Chinese Medicine's institutional research ethics committee. All patients received an explanation of the PVP or PKP procedure and clinical data processing. We also obtained written informed consent from all patients. This investigation complied with the requirements of Transparent Reporting of a Multivariable Prediction Model for Individual Prognosis or Diagnosis. 10 Patients who meet the following criteria were included in our study: (1) complete preoperative basic information, and laboratory and imaging examination of the patients were available, which were re-examined at the prescribed time postoperatively; (2) patients treated with PVP or PKP; (3) patients who presented with severe back pain and limited physical activity, especially when turning over or getting up; (4) patients with a fresh fracture confirmed by magnetic resonance imaging (MRI), who were diagnosed with a single vertebral compression fracture using X-ray and computed tomography; (5) the fracture did not involve the posterior wall of the vertebral body, which was intact; (6) patients with spontaneous fractures or fractures caused by minor trauma; (7) patients who did not develop any infection within 15 cm of the puncture site; (8) patients without cardiopulmonary, liver, and renal failure; and (9) patients without coagulopathy or bleeding tendency. The exclusion criteria for the study were as follows: (1) patients with vertebral compression fracture involving more than two segments; (2) patients with pathological vertebral compression fracture caused by vertebral malignant tumor, metastatic tumor, haemangioma, and so forth; (3) patients with unstable fracture and burst fracture of the posterior wall of the vertebral body; (4) patients with severe spinal degeneration; (5) patients with a history of malignant tumors, dementia, or other mental disorders; (6) patients with insufficient clinical or imaging data; (7) patients with spinal cord compression and obvious neurological symptoms. such as numbness and/or muscle weakness; (8) patients with blood coagulation dysfunction, complications associated with systemic disease, and unable to tolerate surgery; and (9) patients with systemic or local infection. Key researchers extracted the data from the electronic medical records, and image archiving and communication systems.

Percutaneous Kyphoplasty and Vertebroplasty
An experienced chief physician performed all procedures in this study using a unilateral lateral approach to the pedicle of the vertebral arch. The size of the vertebral body and level of compression and vertebral body leakage were used to calculate the dose of bone cement to be injected. The patient was placed in the prone position, with supportive padding to allow the abdomen to hang freely. We used C-arm fluoroscopy to determine and mark the needle entry point, which was located at the lateral side of the pedicle of the vertebral arch on the right side of the vertebral body. After successful local anesthesia, a cement needle was used to puncture the lateral pedicle of the vertebral arch on the right side of the middle and posterior third of the vertebral body under fluoroscopy. The tip of the cement needle puncture was placed 3-4 cm from the spine at an abducent angle of 30 -45 .
Thereafter, the needle core was withdrawn, and the solid cone drill was rotated through the puncture channel to the needle entry site under fluoroscopy. This site was located at the anterior and middle third of the vertebral body under lateral fluoroscopy, and at the spinous process of the vertebral body under anterior fluoroscopy. The solid vertebral drill was withdrawn, followed by the insertion of a balloon catheter for vertebral dilatation to facilitate gradual pressurization and maintenance of the balloon at a certain pressure. The height of the vertebral body was restored partially, and the balloon was pulled out under fluoroscopy. The bone cement was mixed to toothpaste consistency, and slowly injected into the vertebral body via the bone cement filler under fluoroscopy. After the bone cement had hardened, we slowly removed the puncture needle to ensure that there was no leakage of bone cement. Finally, we covered the needle puncture site with sterile patches.
PVP did not entail insertion of a balloon catheter for vertebral dilatation, while the other surgical steps were the same as those for PKP.

Identification Criteria for Outcome Measures
The principal diagnostic criteria of NVCF after PVP/PKP were as follows. (1) After initial pain amelioration, the patient experienced a resurgence of back pain within 2 years of surgery, which was accompanied by discernible local tenderness and limited physical activity, especially when turning over or getting up. (2) The vertebral body exhibited some wedge-shaped changes on X-ray, and MRI revealed low-and high-signal intensities on the T1-and T2-weighted sequences, respectively. These findings confirmed the existence of a NVCF. MRI was also utilized to eliminate the possibility of other spinal disorders, such as infections and malignant tumors.

Selection Criteria for Predictor Variables
We conducted a comprehensive review of the previous literature, and collected and analyzed the postoperative risk factors associated with NVCF, such as laboratory indicators, baseline characteristics, and surgical treatment-related factors. The patients' baseline characteristics included sociodemographic factors [age, sex, time from injury to surgery, history of trauma, body mass index (BMI), bone mineral density (BMD)] and preoperative imaging examination indices (preoperative vertebral height, preoperative compression ratio, average vertebral height, segmental kyphosis angle, thoracolumbar kyphosis Cobb Angle, and coronal Cobb Angle). The surgical treatment factors included fracture site, orthopaedic procedures, and amount of bone cement injected. The laboratory indicators included parathyroid hormone, vitamin D, osteocalcin, β collagen special sequence, total type I collagen amino elongation peptide, red blood cell distribution width, fibrinogen, hemoglobin, activated partial thromboplastin time, homocysteine, alkaline phosphatase, serum Ca, and serum creatinine. Postoperative imaging changes (such as the diffusion of bone cement, vertebral height recovery rate, and whether the bone cement contacted the endplate) were also included in our study. We also investigated whether patients were administered postoperative anti-osteoporosis therapy with zoledronic acid.

Statistical Analysis
Statistical analysis, development, and verification of the model were conducted using SPSS 26.0 statistical software (International Business Machines Corporation, America), R software (The R Foundation for Statistical Computing, Austria, Version 4.2.0; https://www.R-project.org), and RStudio software (Public Benefit Corporation in Delaware, America, Version 2022.07.2-576; https://www.rstudio.com/). Continuous variables were presented as the mean AE SD, and count data were transformed into count classification variables, expressed as ratios. p-values <0.05 were considered statistically significant.
The least absolute shrinkage and selection operator (LASSO) regression method restricts the coefficients of the predictive variables by imposing constraints on the process of parameter estimation, reduces the regression coefficients of independent predictive variables to zero, and minimizes the possibility of over-fitting of the model. 11 We screened the predictability of the ideal risk variables with non-zero coefficients in the LASSO regression model using the LASSO approach. Thereafter, coupled with the characteristics selected in the LASSO regression model, the predictive model was established using multivariate logistic regression analysis. To assess the model's level of calibration, we created a calibration chart for the clinical predictive model. The area under the receiver operating characteristic curve indicates the accuracy with which a model can predict a future event. The net benefit under a tolerable risk threshold consistent with clinical practice was depicted using decision curve analysis, which was employed to assess the model's clinical feasibility. We applied Bootstrap validation resampling (1000 repetitions) to calculate the relative corrected concordance index. The statistical methods used in this study are supported by relevant literature 12,13 and supervised and reviewed by statistical experts at Shandong University of Traditional Chinese Medicine.

Patients' Baseline Characteristics
A total of 200 patients satisfied the eligibility criteria for this study. Participants were divided into the non-NVCF (n = 126) and NVCF groups (n = 74), based on the incidence of NVCF. All patients were monitored for 2 years and included in the final analysis. The laboratory indicators, baseline characteristics, and surgical treatment factors of the two groups are enumerated in Table 1.

LASSO Regression Analysis Results
LOSSA regression analysis was used to estimate the coefficient of each risk factor for NVCF. NVCF was used as a  Figures 2, 3). Fifteen variables were selected for preliminary screening, including time from injury to surgery, osteocalcin, total type I collagen amino elongation peptide, red blood cell distribution width, homocysteine, alkaline phosphatase, BMD, orthopaedic procedure, preoperative compression ratio, thoracolumbar kyphosis Cobb angle, coronal Cobb Angle, whether anti-osteoporosis therapy was administered, diffusion of bone cement, vertebral height recovery rate, and whether the bone cement contacted the endplate. Multivariate logistic regression analysis showed that time from injury to surgery, osteocalcin, homocysteine, BMD, orthopaedic procedure, whether or not out anti-osteoporosis therapy was administered and diffusion of bone cement ( Figure 1) were independent predictors of NVCF ( Table 2).

Development of an Individualized Prediction Model
A model with the above-mentioned seven independent predictors was created for use with the nomogram (Figure 4). We plotted the calibration curve of the predictive model. The AUC of the predictive model was 0.895, demonstrating a high level of predictive accuracy ( Figure 5). The apparent curve of the model fit well with the curve after bias correction, indicating that the predictive model had high discriminatory ability and good fit ( Figure 6). After 1000 Bootstrap resampling verifications, it was confirmed that the concordance index was 0.866, which proved that the discriminatory ability of the model was excellent.

Clinical Utility of the Nomogram for Predicting NVCF Risk
The results of line chart decision curve analysis for factors influencing NVCF showed that this line chart possessed the greatest clinical net benefit for predicting NVCF after bone cement injection if it is within the threshold range of 1% and 91%, as shown in Figure 7.

Discussion
A fter our thorough study and exploration, we finally employed seven easily accessible factors to estimate the likelihood of postoperative NVCF, which would aid clinicians in implementing preoperative and postoperative management of patients in a scientific manner. To the best of our knowledge, our study is the first to combine patients' laboratory indicators with preoperative baseline characteristics and surgical treatment parameters for risk prediction into an easy-to-use nomogram, to evaluate the risk of NVCF in middle-aged and elderly patients with vertebral compression fractures after cement injection.

Effects of Different Surgical Procedure (PKP and PVP) on NVCF
Both PKP and PVP are safe and efficient minimally invasive surgical techniques for the treatment of vertebral compression fractures. Polymethyl methacrylate cement is injected into the fractured vertebral body to achieve mechanical stabilization of the fractured vertebral body, improve the strength of the vertebral body, and relieve pain. 14 However, experts are divided about the procedure with the maximal clinical benefit, especially with respect to the degree of amelioration of postoperative pain symptoms and the probability of NVCF. Griffoni et al. 15 investigated the efficacy and safety of PKP and PVP for vertebral compression fracture using longterm follow-up; they found that both methods resulted in good recovery of vertebral height and improvement of the kyphosis angle, but the risk of adjacent segmental fractures was significantly higher in the PVP group. A network analysis 16 showed that PKP was the best treatment to enhance the quality of life and lower the risk of vertebral re-fracture compared to PVP and conservative treatment, and the risk of adjacent vertebral fracture was substantially lower with PKP than that with PVP. Several studies have linked poor spinal sagittal alignment with a higher risk of vertebral compression fracture in individuals with osteoporosis; moreover, the worse the overall sagittal position, the poorer the quality of life. 17 Cao et al. 18 found that PKP can significantly improve global sagittal plane imbalance caused by vertebral fractures, especially in vertebral fractures involving the thoracolumbar segment, and correct pelvic posterior rotation occurring during sagittal compensatory balance within a short time, which is crucial for realizing the balance between spinal and pelvic (D) The vertebrae in which the bone cement had some contacts with the upper or lower endplate sagittal parameters and maintaining the axis of gravity in the natural position as far as possible. Therefore, these studies provide robust evidence to support screening of risk factors and model development. Additionally, we believe that PKP may be superior to PVP with respect to the lower incidence of postoperative secondary vertebral compression fractures and better quality of life, based on our analysis and previous studies.

Preventive Effects of Osteocalcin on NVCF
Osteocalcin, a bone biochemical marker secreted by osteoblasts, is the most abundant non-collagenous protein in bone, which is essential for the synthesis and turnover of bone matrix. 19,20 The plasma level of osteocalcin is thought to reflect the degree of bone formation. 21 Osteocalcin, in conjunction with other bone biochemical markers, can be used to quantitatively monitor the level of bone metabolism. Utilizing these quantitative changes for risk prediction may bolster timely and comprehensive evaluation during the bone healing phase and predict the risk of complications and the degree of damage to fracture healing. 22 Poundarik et al. 23 found that dilatational bands formed by the osteocalcinosteopontin interaction are vital for energy dissipation and  maintaining the fracture toughness of bone. Moreover, empirical evidence supports the significant role of osteocalcin in bone fracture resistance. Nikel et al. 24 were the first to prove the importance of osteocalcin and osteopontin in creep and fatigue of bone, and demonstrated that their deficiency seriously compromised the plasticity of the bone matrix and its ability to withstand periodic load. Therefore, osteocalcin plays a key role in improving the toughness and anti-fracture properties of bone, which has a far-reaching impact on the prevention of NVCF after bone cement injection.

Effects of Excessive Homocysteine on NVCF
Numerous experimental studies have shown that homocysteine is an independent risk factor for vertebral compression fractures in the elderly; elevated homocysteine levels increase the risk of fractures in humans. [25][26][27] Homocysteine can reduce bone vascular flow, which can destroy the crosslinking of collagen molecules and affect the normal calcification process. 28,29 Additionally, excessive homocysteine can induce apoptosis of osteoblasts, osteocytes, and human bone marrow stromal cells by enhancing the differentiation of osteoclasts, resulting in damage to the mechanical strength of bone. 28,29 This exerts adverse effects on the healing of fracture sites, balance of bone metabolism, and the recovery of bone mechanical properties after bone cement injection. Substantial evidence indicates that the decline in homocysteine levels can reduce the risk of cardiovascular and cerebrovascular diseases such as stroke, delay the decline of coordination function, increase bone strength in the elderly, and reduce the risk of fall secondary to skeletal muscle weakness, thus reducing the risk of vertebral compression fracture in the elderly. [30][31][32] According to a Japanese study, homocysteine was an independent risk factor for moderate and severe vertebral compression fractures; the risk of subsequent vertebral fractures and other osteoporotic fractures was considerably higher in patients with severe fractures than that in patients with moderate fractures. 33,34 According to the above-mentioned study, homocysteine elevation may have an adverse effect on the patients' ability to recover after bone cement injection and increase the risk of NVCF.

Effects of Surgical Treatment and Anti-Osteoporosis Treatment on NVCF
Osteoporosis can lead to the destruction of the microstructure of the human bone and decrease in bone strength. 35 Patients with osteoporosis are susceptible to compression fracture and other fragility fractures, while patients with vertebral compression fractures may experience serious physical limitations, such as back pain and dysfunction. 36 If not treated as soon as possible, the height of the vertebral body may collapse further with the increase in the fracture age and range of motion of the fracture site, and the repeated compressive load at the fracture site may lead to bone loss near the endplate, which may increase the risk of NVCF. 37 Clinical studies with long-term follow-up have shown that regular administration of anti-osteoporotic drugs, such as teriparatide or zoledronic acid, to post-PKP patients with osteoporotic vertebral compression fractures can significantly reduce the incidence of NVCF and back pain, increase BMD, effectively restore postoperative function, and improve patients' quality of life, consistent with the present study. 38,39 Therefore, we believe that early treatment of fractures and targeted anti-osteoporosis treatment play a key role in the prevention of NVCF.
Several studies have focused on the degree of diffusion of bone cement, which exerts a significant impact on biomechanical aspects, such as the stability and mechanical properties of the fracture site. Some studies have shown that the asymmetric distribution of the vertebral body and excessive filling of cement volume lead to the transfer and switching of unilateral load, resulting in NVCF. 40,41 The height and strength of the vertebral body recovered well, and the risk of vertebral recompression decreased significantly when the bone cement was in contact with the upper and lower endplates. 40,41 At the same time, some experts believe that when the cement distribution is wide enough (i.e., the area between the midline and contralateral pedicle), unilateral PKP can significantly restore the height of the vertebral body and provide a good therapeutic effect. 15 Therefore, good cement diffusion has a positive impact on fracture site stability, mechanical properties, and stress distribution, effectively preventing NVCF in the vertebral body. Clinical Significance of the Study Although previous studies have conducted in-depth and extensive investigations into the effects of osteocalcin and homocysteine on vertebral compression fractures and bone metabolic balance in middle-aged and elderly people, no study has investigated their association with the incidence of NVCF after bone cement injection. In the current study, osteocalcin and homocysteine were combined with preoperative baseline characteristics and surgical treatment factors to comprehensively study the risk of NVCF. Notably, no study has comprehensively investigated the risk of NVCF by combining osteocalcin and homocysteine with preoperative baseline characteristics and surgical treatment factors as risk factors. By incorporating all relevant risk factors, our study provides a comprehensive, accurate, and easy-toimplement nomogram to calculate the probability of NVCF in patients after bone cement injection surgery, which may have profound implications for postoperative prevention, treatment, and targeted follow-up, and also serve as a roadmap for future studies.

Strengths and Limitations
Nomograms are a scientific tool with high accuracy and provides us with easy-to-understand prognostic models. Our strength is to take advantage of nomograms and comprehensively combine patients' laboratory indicators with preoperative baseline characteristics and surgical treatment parameters to produce a risk prediction model which is conducive to clinical decision-making with high credibility and easy operation. This study is beset by several limitations. First, although our nomogram was robust and underwent thorough internal verification using Bootstrap sampling, the clinical data were acquired from our hospital's orthopaedic department. Hence, the lack of comparison with data from other centres, and lack of external verification, necessitate external evaluation in other middle-aged and elderly populations with vertebral fractures from more countries and regions. Although we comprehensively collected the clinical data from our sample, another limitation is the small sample size, and selection bias resulting from the retrospective study design. Therefore, it is necessary to perform collaborative multi-center studies to further augment our data and prospective research to verify the accuracy of this nomogram. The current analysis investigated the influence of osteocalcin and homocysteine on NVCF after bone cement injection only from the macrostructural perspective. Further cellular and molecular studies are required to elucidate the biological mechanisms associated with osteocalcin and homocysteine and fracture susceptibility or factors leading to new fractures after surgery, to verify the model's correctness from a microscopic perspective. The ROC curve reflects the curve between sensitivity and specificity. The X axis is specific, also known as the false positive rate, the closer the X axis is to zero, the higher the accuracy; the larger the Y axis, the better the accuracy; the area under the curve (AUC) indicates the accuracy of prediction, and the higher the AUC value, the larger the area under the curve, the higher the accuracy of prediction Actual Probability Mean absolute error=0.027 n=200 B= 1000 repetitions, boot Apparent Bias-corrected Ideal Fig. 6 The fitting was repeated 1000 times, and the average absolute error was 0.027 (n = 200). The X axis represents the predicted NVCF risk; the Y axis represents the actual risk of NVCF; the diagonal dashed line represents the perfect prediction of the ideal model; and the solid line represents the performance of the line chart, which is closer to the diagonal dashed line to indicate a better prediction

Conclusion
Time from injury to surgery more than 7 days, osteocalcin levels below normal, homocysteine levels above the normal range, presence of osteoporosis, mode of operation, that is, percutaneous vertebroplasty, lack of postoperative antiosteoporosis treatment, and poor diffusion of bone cement are independent risk factors of NVCF in middle-aged and elderly patients with vertebral compression fracture after bone cement injection. The nomogram based on these seven factors can objectively and accurately prognosticate the likelihood of NVCF in middle-aged and elderly patients with vertebral compression fracture following bone cement injection and assist clinicians in timely determination of the risk of clinical events and development of individualized treatment plans. Future large-scale studies are warranted to determine the predictive efficiency of the model and explore whether targeted postoperative patient management using the model's predictions can reduce the probability of NVCF.

Author Contributions
A ll authors had full access to the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.  Fig. 7 Decision curve analysis. The y-axis represents net income. The blue curve represents the NVCF risk nomogram. The horizontal solid line represents the assumption that no patient has developed NVCF.
The thin dotted line represents the hypothesis that NVCF occurs in all patients