INTRODUCTION
With the increasing global aging population, osteoporosis has become the most prevalent bone disease, with an estimated worldwide prevalence of 18.3%, notably higher among women [ 1 , 2 ]. Osteoporotic vertebral compression fracture (OVCF) is one of the most common complications associated with osteoporosis, and its incidence continues to rise [ 3 , 4 ]. OVCF is known to cause persistent pain and kyphotic deformity, restrict mobility, and significantly impair quality of life, potentially leading to increased mortality [ 5 ]. Percutaneous vertebral augmentation (PVA), which includes percutaneous vertebroplasty and percutaneous kyphoplasty, is a widely used and effective minimally invasive surgical treatment for OVCF. This procedure can partially restore vertebral height, reconstruct spinal stability, provide rapid pain relief, and improve patients’ quality of life [ 6 , 7 ].
Although PVA is widely regarded as a safe surgical procedure, it is still associated with certain complications, including cement leakage, cement embolism, nerve injury, epidural hematoma, and infection [ 8 ]. Notably, while the incidence of spinal infection after vertebral augmentation (SIAVA) is exceptionally low, its occurrence necessitates prolonged administration of high-dose antibiotics or additional surgical intervention and is associated with high rates of disability and mortality [ 9 , 10 ]. Although sharing standard features with other postoperative spinal infections, SIAVA may possess a distinct pathophysiology due to the unique context of PVA, which is a minimally invasive procedure involving cement implantation in often elderly, frail patients with osteoporosis. This suggests the possibility of specific, procedurerelated risk factors warranting dedicated investigation. However, existing research on SIAVA primarily consists of case reports, with no comprehensive analysis of related risk factors. Robust evidence from large-scale, well-designed studies on its precise incidence and independent risk factors remains limited. Furthermore, potential risk factors unique to the PVA procedure, such as specific cement leakage patterns or preoperative imaging findings indicative of vertebral viability, have not been systematically evaluated in large cohorts.
To address these gaps, we conducted a multicenter, case-control study to quantify the incidence of SIAVA and identify its independent risk factors, with a particular focus on imaging characteristics, procedure-specific factors, and patient comorbidities. By employing propensity score matching (PSM) and multivariate regression, this study provides critical data to inform clinical risk stratification and prevention strategies.
MATERIALS AND METHODS
1. Study Design and Patient Selection
This retrospective case-control study was conducted at 4 medical centers in China between January 1, 2015, and December 31, 2021. Consecutive patients who underwent PVA for OVCFs and subsequently developed a spinal infection were identified as cases. For each case, 3 control patients were selected from the noninfected cohort using PSM based on key demographic and clinical variables. The inclusion criterion required a minimum follow-up duration of one year. To ensure cohort homogeneity, the following patients were excluded: (1) those with preoperative pathologies such as spinal metastases, multiple myeloma, or vertebral hemangioma; (2) those with incomplete clinical or imaging data; (3) those with preoperative clinical, laboratory, or imaging evidence suggestive of preexisting spinal infection, including typical imaging findings of spinal infection or markedly elevated inflammatory markers without other identifiable causes; and (4) those with less than one year of follow-up.
Ethical approval for this study was obtained from the ethics committee of Guangdong Province Hospital of Chinese Medicine (approval number: ZE2023-176-01), and the research was conducted in accordance with the World Medical Association Declaration of Helsinki. Owing to the retrospective study design, the ethics committee waived the requirement for written informed consent.
2. SIAVA Diagnosis and Management
The diagnosis of SIAVA was established according to a 2-tiered diagnostic framework adapted from the Centers for Disease Control and Prevention criteria and published literature specific to spinal infections following vertebral augmentation [ 11 - 15 ]. Given the retrospective nature of this study and the inherent limitations of relying solely on microbiological confirmation in all cases, we adopted a hierarchical diagnostic approach to balance methodological rigor with clinical realism.
3. Definite SIAVA (Microbiologically Confirmed)
A patient was classified as having definite SIAVA if pathogenic bacteria were identified through percutaneous biopsy or intraoperative sampling, with or without corresponding clinical, laboratory, or imaging manifestations [ 11 ].
4. Probable SIAVA (Clinically Diagnosed Without Microbiological Confirmation)
In the absence of microbiological confirmation, a patient was classified as having probable SIAVA only if all 3 of the following criteria were met simultaneously: (1) Persistent or new-onset back pain, localized tenderness, restricted mobility, or wound drainage, with or without fever (>38°C) [ 12 , 13 ]; (2) Elevated inflammatory markers, specifically a C-reactive protein (CRP) level >10 mg/L and/or an erythrocyte sedimentation rate (ESR) >30 mm/hr, in the absence of other identifiable causes [ 14 ]; (3) Typical imaging findings of spinal infection, including bone destruction centered around the bone cement, abscess formation, or definite rim enhancement on contrast-enhanced MRI, with explicit exclusion of alternative diagnoses such as cementrelated inflammatory reaction, fracture nonunion, or degenerative endplate changes [ 13 - 15 ].
To ensure methodological consistency across all centers in this retrospective analysis, a uniform set of diagnostic and management criteria was applied to patients diagnosed with SIAVA. Following readmission, each patient’s evaluation adhered to this protocol, which mandated standardized laboratory testing of inflammatory markers and comprehensive spinal imaging, with all studies interpreted according to the predefined criteria. Furthermore, a definitive microbiological diagnosis was performed in every case through percutaneous or intraoperative biopsy; the obtained tissue samples were processed consistently across sites for microbial culture, histopathology, and metagenomic next-generation sequencing. To enhance data reliability, all imaging studies, laboratory results, and biopsy reports were independently reviewed by 2 senior spine surgeons blinded to patient outcomes, with any discrepancies resolved by consensus. The administered treatment strategy (antibiotic therapy alone or revision surgery) was systematically documented. Functional outcomes were formally evaluated at the final follow-up. This consistent application of criteria ensured that the primary outcome and key exposure variables were determined using comparable, high-quality data throughout the study cohort.
5. Data Collection
Data on patient-specific and procedure-specific variables were extracted from electronic medical records at all participating centers using a standardized data collection form, developed based on clinical expertise and previous literature [ 10 , 14 , 16 , 17 ]. The collected variables included demographic information (age, sex, height, weight, and body mass index [BMI]), clinical characteristics (history of trauma, disease duration, level of fractured vertebrae, long-term preoperative steroid use, and operative duration), preoperative comorbidities (pulmonary infection, chronic obstructive pulmonary disease [COPD], urinary tract infection, hypertension, and type 2 diabetes), preoperative laboratory values (hemoglobin, ESR, CRP, and albumin levels, with serum albumin [Alb] ≤35 g/L defined as hypoalbuminemia) [ 18 ], and imaging features (presence of an intravertebral fluid sign on preoperative MRI, and presence and type of postoperative bone cement leakage).
All data were anonymized and independently assessed by 2 senior spine surgeons. Interobserver agreement was excellent for both the detection of the intravertebral fluid sign (κ=0.88) and the classification of cement leakage type (κ=0.85). Any discrepancies were resolved by consensus to finalize the dataset for analysis.
The fluid sign in the fractured vertebral body was defined as the presence of an intravertebral vacuum cleft with fluid filling as the predominant feature. Based on preoperative MRI data, the typical appearance consisted of a low signal in T1-weighted sequences and a high signal in T2-weighted sequences within the intravertebral cleft [ 19 ]. The presence of bone cement leakage was assessed on postoperative anteroposterior and lateral x-ray images. The type of bone cement leakage was classified according to the method described by Tomé-Bermejo et al. [ 20 ]
6. Statistical Analyses
This case-control study designated patients diagnosed with SIAVA as cases and those without infection as controls. To minimize confounding bias, 1:3 PSM was performed using the nearest-neighbor method with a caliper width of 0.2. Matching covariates included age, sex, height, weight, BMI, history of trauma, and the number of fractured vertebrae. These covariates were selected a priori as fundamental demographic and injuryrelated factors that could significantly influence both the baseline risk of infection and the decision to operate, thus requiring balancing to isolate the effect of other putative risk factors. The balance of baseline characteristics between groups following matching was assessed using standardized mean differences (SMDs).
Data with a normal distribution are presented as mean±standard deviation and were compared using an independentsamples t-test. Nonnormally distributed data are presented as median (interquartile range [IQR]) and were compared using the Mann-Whitney U-test. Categorical variables are presented as frequency (percentage) and were compared using the chisquare test or Fisher exact test, as appropriate.
After matching, conditional logistic regression was used to account for the matched pair structure, thereby preserving the pairing information and reducing potential bias. Candidate variables for least absolute shrinkage and selection operator (LASSO) regression were preselected based on clinical expertise and previously reported risk factors in the literature [ 21 ]. A total of 16 candidate variables were entered into the LASSO regression, including disease duration, hypertension, diabetes mellitus, COPD, long-term corticosteroid use, preoperative urinary tract infection, albumin ≤35 g/L, CRP, ESR, intravertebral fluid sign, pulmonary infection, operative duration, overall cement leakage, type B leakage, type C leakage, and type D leakage. Tenfold cross-validation was applied, and the more conservative lambda.1se criterion was used to select the final model variables. Variables retained by LASSO were included in the final multivariable logistic regression model, and odds ratios (ORs) and 95% confidence intervals (CIs) were reported.
Model discrimination was assessed using the area under the receiver operating characteristic curve (AUC). Internal validation was performed using bootstrap resampling with 200 replicates, and the optimism-corrected AUC was calculated to evaluate model stability and potential overfitting. Given the limited number of infection events (n=42), Firth penalized conditional logistic regression was additionally performed as a sensitivity analysis to reduce small-sample bias and assess the robustness of the findings. There were no missing data for key variables in the matched cohort; therefore, complete-case analysis was performed (all cases had complete data). All statistical tests were 2-tailed, and a p-value<0.05 was considered statistically significant. Statistical analyses were performed using IBM SPSS Statistics ver. 25.0 (IBM Co., USA) and R ver. 4.4.2 (R Foundation for Statistical Computing, Austria).
RESULTS
1. Patient Cohort and PSM
During the study period, 7,797 patients who underwent PVA were identified, of whom 42 developed SIAVA, yielding an incidence rate of 0.54%. The infection rates across the 4 centers were as follows: center 1, 0.76% (20 of 2,624); center 2, 0.20% (4 of 1,963); center 3, 0.79% (16 of 2,014); and center 4, 0.17% (2 of 1,196). The median time from PVA to SIAVA diagnosis was 4.3 (IQR, 2.8–11.7) weeks. PSM yielded a well-balanced cohort of 168 patients (42 cases and 126 controls) for subsequent analysis.
Before matching, significant baseline differences were observed between the SIAVA cases and the overall noninfected cohort (n=7,755), particularly in sex distribution (42.9% vs. 18.2% male, p<0.001) and history of trauma (57.1% vs. 41.6%, p=0.041). Following 1:3 matching based on age, sex, height, weight, BMI, trauma history, and number of fractured vertebrae, all baseline characteristics were comparable between groups, with no statistically significant differences (all p>0.05), confirming the effectiveness of the matching procedure ( Table 1 ). After matching, all SMDs for the matched covariates were less than 0.2, indicating adequate balance between the 2 groups.
2. Microbiological Profile and Clinical Management of SIAVA
Among the 42 patients with SIAVA, 26 (61.9%) had positive microbiological cultures. These 26 cases were classified as definite SIAVA, whereas the remaining 16 culture-negative cases were classified as probable SIAVA based on clinical and imaging criteria. The distribution of specific pathogens is summarized in Table 2 . Gram-positive bacteria predominated, with Staphylococcus epidermidis being the most frequently isolated organism (8 of 26, 30.8%), followed by Staphylococcus aureus (5 of 26, 19.2%).
Regarding therapeutic management, 25 patients (59.5%) underwent revision surgery, while the remaining 17 (40.5%) received antibiotic therapy alone. Among the 26 culture-positive patients, 14 (53.8%) underwent revision surgery, and 12 (46.2%) received conservative treatment; among the 16 culture-negative patients, 11 (68.8%) underwent revision surgery, and 5 (31.3%) received conservative treatment.
At the final follow-up, functional outcomes showed that 23 patients (54.8%) regained independent ambulation; 11 patients (26.2%) required walker assistance, and 6 patients (14.3%) became wheelchair-dependent. Two deaths (4.8%) occurred during the follow-up period.
3. Comparison of Clinical Characteristics
Comparative analysis of the 168 matched patients revealed significant differences in clinical characteristics between the 2 groups ( Table 3 ). The infection group had significantly higher proportions of patients with hypertension (59.5% vs. 38.9%, p=0.020), diabetes mellitus (31.0% vs. 12.7%, p=0.009), pulmonary infection (31.0% vs. 9.5%, p=0.001), and COPD (28.6% vs. 11.9%, p=0.013).
Significant differences were also observed in laboratory and imaging findings. The infection group showed a higher prevalence of hypoalbuminemia (Alb ≤35 g/L: 47.6% vs. 11.9%, p< 0.001), anemia (61.9% vs. 27.8%, p<0.001), presence of an intravertebral fluid sign on MRI (78.6% vs. 43.7%, p<0.001), and type D cement leakage (40.5% vs. 15.9%, p=0.001). Additionally, operative duration was significantly longer (p<0.001), and ESR levels were higher (p=0.012) in the infection group compared with controls.
4. Risk Factor Analysis
After LASSO variable selection with 10-fold cross-validation and the lambda.1se criterion ( Fig. 1 ), 4 variables were retained for the final multivariable model: pulmonary infection, intravertebral fluid sign, type D cement leakage, and Alb ≤35 g/L.
The final multivariable conditional logistic regression model ( Table 4 ) showed that all 4 variables were independently associated with SIAVA. The strongest association was observed for intravertebral fluid sign (OR, 6.17; 95% CI, 1.92–19.9; p=0.002), followed by albumin ≤35 g/L (OR, 3.29; 95% CI, 1.15–9.45; p=0.027), pulmonary infection (OR, 3.64; 95% CI, 1.21–11.0; p=0.022), and type D cement leakage (OR, 2.93; 95% CI, 1.10– 7.85; p=0.032). The model demonstrated good discrimination, with an apparent AUC of 0.866 and an optimism-corrected AUC of 0.782 after bootstrap internal validation (200 resamples) ( Supplementary Table 1 ), indicating acceptable model stability given the limited sample size. Sensitivity analysis using Firth penalized conditional logistic regression yielded results consistent with the main analysis ( Table 5 , left column), with all 4 variables remaining statistically significant (p<0.05), further supporting the robustness of the findings.
A sensitivity analysis restricted to the 26 culture-confirmed cases ( Table 5 , right column) confirmed the independent association for intravertebral fluid sign (OR, 8.08; 95% CI, 2.04–32.0; p=0.003) and Alb ≤35 g/L (OR, 3.86; 95% CI, 1.03–14.4; p=0.044). The associations for type D leakage (OR, 2.77; 95% CI, 0.75–10.2; p=0.127) and pulmonary infection (OR, 4.15; 95% CI, 0.95–18.1; p=0.059) were not statistically significant.
DISCUSSION
As a well-established treatment for OVCF, PVA carries a low but clinically significant risk of SIAVA. Although rare, SIAVA can result in severe complications, including spinal instability, neurological deficits, and even mortality [ 9 , 10 , 13 , 14 ]. This multicenter, retrospective case-control study systematically examined the incidence and risk factors of SIAVA. The observed incidence rate of 0.54% aligns with the previously reported range of 0.32% to 1.9% [ 9 10 , 13 , 14 , 17 ]. Using LASSO regression followed by conditional logistic regression, we identified 4 independent risk factors: pulmonary infection, intravertebral fluid sign on MRI, type D cement leakage, and Alb ≤35 g/L. The model demonstrated good discrimination, with an optimism‑corrected AUC of 0.782 after bootstrap internal validation. The robustness of these findings was further supported by Firth penalized conditional logistic regression, which confirmed the independent association of all 4 risk factors after accounting for small-sample bias.
Notably, the infection-related mortality rate in this cohort (4.8%) was lower than the 15.1% reported by Park et al. [ 17 ], possibly reflecting recent advancements in early diagnosis and comprehensive treatment strategies. Despite this, the clinical burden associated with functional impairment and the need for reoperation remains considerable.
1. Hypoalbuminemia and Nutritional Status
Hypoalbuminemia has been reported as a significant risk factor for surgical site infection following PVA [ 10 , 13 , 17 ]. In the present study, multivariable analysis further confirmed that preoperative hypoalbuminemia is an independent risk factor for SIAVA, highlighting the critical role of nutritional status in infection prevention. The association between hypoalbuminemia and postoperative infection is well supported by extensive literature [ 22 - 24 ]. Malnutrition leads to widespread impairment of immune function, including reduced neutrophil chemotaxis, suppressed antibacterial activity, and diminished synthesis of key immune components [ 25 ]. In addition, low serum albumin levels affect colloid osmotic pressure, contributing to tissue edema and thereby creating a microenvironment conducive to bacterial proliferation [ 26 ].
Notably, the pathogenic spectrum identified in this study was predominantly composed of low-virulence organisms, such as S. epidermidis and other coagulase-negative staphylococci, consistent with previous reports [ 10 , 17 ]. These organisms, although typically considered less aggressive, may cause significant clinical morbidity in immunocompromised patients.
Based on these findings, we recommend incorporating hypoalbuminemia (Alb ≤35 g/L) as a key indicator in preoperative risk assessment for PVA. Active nutritional support may serve as an adjunctive strategy to reduce the incidence of SIAVA. How-ever, prospective studies are needed to confirm its efficacy.
2. Preoperative Pulmonary Infection
Preoperative pulmonary infection was identified as an independent risk factor for SIAVA in the main analysis (OR, 3.64; p=0.022). The mechanism may involve hematogenous spread, as active pulmonary infection could serve as a distant focus that disseminates pathogens to the surgical site via the bloodstream [ 13 , 14 ]. As early as 1983, Weir and Muraleedharan27 proposed the concept of locus minoris resistentiae, indicating that sites of injury in the body are more susceptible to pathogen invasion. Active pulmonary infection, as a distant focus, may disseminate pathogens to the surgical site via the bloodstream; however, the exact route of dissemination remains unclear. In cases of pyogenic spondylodiscitis, arterial spread is considered a primary pathway [ 28 ]. A single segmental artery often supplies parts of 2 adjacent vertebral bodies and the intervening disc, which may explain the frequent involvement of consecutive segments. Additionally, in the lumbar region, the Batson vertebral venous plexus may play a role. This valveless, bidirectionalflow system provides a potential route for retrograde dissemination of pathogens [ 29 ].
In the sensitivity analysis restricted to the 26 culture-confirmed cases, the association for pulmonary infection was attenuated and no longer statistically significant (OR, 4.15; p=0.059), likely due to reduced sample size and consequent loss of statistical power. Nevertheless, the OR remained >4, suggesting a potentially meaningful clinical association. Therefore, elective PVA should be postponed in patients with active pulmonary or other systemic infections until the primary infection is adequately controlled, as this may help reduce the risk of SIAVA.
3. Procedure- and Imaging-Specific Risk Factors
Interestingly, to our knowledge, this is the first report identifying type D bone cement leakage (intradiscal leakage) as an independent risk factor for SIAVA. This association remained robust after LASSO variable selection and bootstrap internal validation. In the sensitivity analysis, the association was not statistically significant (OR, 2.77; p=0.127), likely due to reduced sample size; nevertheless, the OR remained above 2, and the direction and magnitude were consistent with the main analysis.
Polymethyl methacrylate (PMMA), the material of choice for its stability [ 30 ], is nondegradable and may provoke a chronic local inflammatory response upon leakage [ 31 ]. This inflammation can induce pericemental and discal signal changes on imaging, mimicking an actual infection and leading to diagnostic confusion, as documented in prior reports [ 32 ]. Most spinal infections are hematogenous, initially seeding the vascular vertebral body. The adult intervertebral disc, however, is largely avascular, with bacteria typically accessing it via the adjacent endplate [ 33 ]. We hypothesize that cement leakage into the disc could theoretically alter this anatomy by creating an intradiscal cement bridge. Although direct evidence is lacking, this aberrant structure might serve as a potential conduit, facilitating the spread of infection from the primary vertebral nidus into the disc and, in theory, to adjacent vertebrae, thereby expanding the infectious focus.
In our cohort, this mechanism is illustrated by a patient with preoperative pulmonary infection and intradiscal leakage, in whom the infection subsequently extended to the disc and an adjacent vertebra ( Fig. 2 ). Furthermore, the poor vascularity within the crevices formed by cement leakage likely impedes antibiotic penetration, creating a niche favorable for bacterial colonization and persistence, thereby making eradication of infection particularly challenging. Therefore, perfecting surgical technique to prevent intradiscal leakage is a critical modifiable factor for risk reduction. This technical vigilance is especially warranted in patients considered at higher baseline risk for hematogenous seeding.
Histologically, a compression fracture with an intravertebral cleft sign closely resembles vertebral osteonecrosis, and the latestage cleft may be replaced by fluid. As a result, T2-weighted MRI sequences typically show high signal, with an 81% probability that the intravertebral vacuum cleft in the thoracolumbar segment is filled with liquid [ 34 ]. Although intravertebral gas or fluid accumulation is uncommon in spinal infection, its presence cannot be entirely excluded [ 35 ]. In fact, active tissue inflammation during infection promotes fluid accumulation and edema, manifesting as the “fluid sign” ( Fig. 3 ). In our study, this finding remained significant in both the main and sensitivity analyses, confirming its robustness.
Nevertheless, although the intravertebral fluid sign was statistically associated with SIAVA, it may also reflect preexisting vertebral necrosis, nonunion, or even undetected infection [ 36 ]. We hypothesize that MRI findings of intravertebral fluid and computed tomography evidence of fissures or cavities may indicate local vascular injury and necrosis, potentially compromising blood supply and creating an environment conducive to bacterial seeding. Furthermore, incomplete filling of these cavities with bone cement may theoretically result in mechanical instability within the vertebral body. This instability, when combined with the inherent stress concentration at the thoracolumbar junction, may further compromise spinal stability and potentially exacerbate an infectious process.
4. Clinical Outcomes and Integrated Prevention Strategies
The long-term functional outcomes in our SIAVA cohort were sobering. At final follow-up, only 54.8% (23 of 42) of patients achieved independent ambulation, while 40.5% (17 of 42) required a walker or became wheelchair-dependent. The mortality rate was 4.8% (2 of 42). These stark outcomes highlight the critical need for proactive risk management.
Based on the independent risk factors identified in this study, we propose the following considerations for perioperative management: (1) Active pulmonary or other systemic infections should be adequately treated before elective PVA, given their potential for hematogenous seeding and borderline association with SIAVA observed in this study. (2) Nutritional support for patients with hypoalbuminemia (Alb ≤35 g/L). (3) When an intravertebral fluid sign is present on MRI, a lower threshold for intraoperative biopsy may be warranted for microbiological diagnosis. (4) Meticulous surgical technique should be employed to minimize cement leakage, particularly type D (intradiscal) leakage, which was identified as a significant risk factor.
5. Limitations
This study has several limitations. First, the retrospective design may introduce information bias. PSM was restricted to demographic and injury-related variables; other important confounders, such as frailty, surgical nuances, immunological status, and center-level variations, could not be matched, and residual confounding may persist. Second, given the rarity of SIAVA, the limited number of cases (n=42) reduced statistical power. LASSO and bootstrap internal validation helped mitigate overfitting. In the sensitivity analysis restricted to 26 culture-confirmed cases, the associations for type D leakage (OR, 2.77) and pulmonary infection (OR, 4.15) were not statistically significant, likely due to reduced power; however, their ORs remained above 2, suggesting potentially meaningful associations. External validation in independent cohorts is warranted. Third, the inclusion of 16 probable (culture-negative) SIAVA cases may have introduced diagnostic heterogeneity. Moreover, the intravertebral fluid sign, though statistically associated with SIAVA, may also reflect preexisting vertebral necrosis, nonunion, or undetected subclinical infection, and should be interpreted with caution. Fourth, although this was a 4-center study, the number of infection events in each center was too small to permit reliable center-level heterogeneity analysis or adjustment for center effects. Potential residual confounding due to center-level differences (e.g., surgical technique, perioperative management, infection control) was therefore unavoidable. Future large-sample, prospective studies are needed to validate our findings across different medical centers.
CONCLUSION
This multicenter case-control study of 7,797 patients reports an incidence of 0.54% for SIAVA. Multivariable analysis identified 4 independent risk factors: preoperative pulmonary infection, intravertebral fluid sign on MRI, type D cement leakage, and Alb ≤35 g/L. To our knowledge, this is the first study to identify intravertebral fluid sign on MRI and type D (intradiscal) cement leakage as independent risk factors for SIAVA. The model showed acceptable discrimination (optimism-corrected AUC=0.782).
Based on these findings, we recommend preoperative control of active pulmonary infection, nutritional support for patients with hypoalbuminemia, a lower threshold for intraoperative biopsy when an MRI intravertebral fluid sign is present, and meticulous surgical technique to prevent type D cement leakage. Future prospective studies are warranted to validate these risk factors.
Conflict of Interest
The authors have nothing to disclose.
Funding/Support
This work was supported by grants from Youth S&T Talent Support Programme of GDSTA (SKXRC 2025108), the Science and Technology Program for Social Development of Zhuhai City (2420004000303), Li Yong Guangdong Provincial Famous TCM Master Inheritance Studio Guangdong and Province Fourth Batch of Distinguished Traditional Chinese Medicine Practitioners Mentorship Program (Document No. 108 [2023], Guangdong TCM Administration).
Author Contribution
Conceptualization: RC, MC, YL; Data curation: TX, XFeng, LY, HY, XL, SL, MY, CP; Formal analysis: RC, XFeng; Funding acquisition: RC, YL; Methodology: RC, JC; Project administration: YL; Visualization: RC, TX, XFeng, YL; Writing – original draft: RC, JC, XL, SL, MY, CP; Writing – review & editing: RC, TX, XFeng, LY, HY, XFang, MC, YL.
Supplementary Material
Supplementary Table 1 is available at https://doi.org/10.14245/ns.26520616.0308 .
Model predictive performance and bootstrap validation
REFERENCES
1 Walker MD Shane E Postmenopausal osteoporosis N Engl J Med 2023 389 1979 91 37991856 10.1056/NEJMcp2307353
2 Xiao PL Cui AY Hsu CJ Global, regional prevalence, and risk factors of osteoporosis according to the World Health Organization diagnostic criteria: a systematic review and meta-analysis Osteoporos Int 2022 33 2137 53 35687123 10.1007/s00198-022-06454-3
3 Zhang B Qin Y Yu J Global research landscape of osteoporotic vertebral compression fractures: a 20-year bibliometric and visualization study Asian J Surg 2026 49 1470 1
4 Dai C Liang G Zhang Y Risk factors of vertebral refracture after PVP or PKP for osteoporotic vertebral compression fractures, especially in Eastern Asia: a systematic review and meta-analysis J Orthop Surg Res 2022 17 161 35279177 10.1186/s13018-022-03038-z PMC8917756
5 Park JS Park YS Survival analysis and risk factors of new vertebral fracture after vertebroplasty for osteoporotic vertebral compression fracture Spine J 2021 21 1355 61 33971326 10.1016/j.spinee.2021.04.022
6 Xie LL Yu H Yuan X A review of the zoledronic acid combined with percutaneous vertebroplasty in the osteoporotic vertebral compression fracture Asian J Surg 2023 46 642 4 35871969 10.1016/j.asjsur.2022.07.020
7 Liu D Wen T Li X Percutaneous vertebroplasty versus balloon kyphoplasty in the treatment of osteoporotic vertebral compression fractures: evaluating the overlapping metaanalyses Pain Physician 2024 27 E383 94 38805534
8 Cavka M Delimar D Rezan R Complications of percutaneous vertebroplasty: a pictorial review Medicina (Kaunas) 2023 59 1536 37763655 10.3390/medicina59091536 PMC10533082
9 Liao JC Lai PL Chen LH Surgical outcomes of infectious spondylitis after vertebroplasty, and comparisons between pyogenic and tuberculosis BMC Infect Dis 2018 18 555 30419832 10.1186/s12879-018-3486-x PMC6233540
10 Abdelrahman H Siam AE Shawky A Infection after vertebroplasty or kyphoplasty. a series of nine cases and review of literature Spine J 2013 13 1809 17 23880354 10.1016/j.spinee.2013.05.053
11 Li Z Zhang Q Lian X Value of metagenomic nextgeneration sequencing in the diagnosis of native pyogenic spinal infections: a multicenter, retrospective observational study Spine J 2025 25 640 8 39615696 10.1016/j.spinee.2024.11.004
12 Mangram AJ Horan TC Pearson ML Guideline for prevention of surgical site infection, 1999. Centers for Disease Control and Prevention (CDC) Hospital Infection Control Practices Advisory Committee Am J Infect Control 1999 27 97 132 quiz 133-4; discussion 96 10196487
13 He Y Ke Z Advances in the diagnosis and treatment of spinal infection after vertebral augmentation Adv Clin Med 2025 15 523 30
14 Yuan H Li Y Wang J Spinal infection after vertebral augmentation: a covert complication with serious havoc Arch Orthop Trauma Surg 2024 144 1461 71 38273125 10.1007/s00402-024-05205-9
15 Yoon YK Jo YM Kwon HH Differential diagnosis between tuberculous spondylodiscitis and pyogenic spontaneous spondylodiscitis: a multicenter descriptive and comparative study Spine J 2015 15 1764 71 25862505 10.1016/j.spinee.2015.04.006
16 Vats HS McKiernan FE Infected vertebroplasty: case report and review of literature Spine (Phila Pa 1976) 2006 31 E859 62 17047535 10.1097/01.brs.0000240665.56414.88
17 Park JW Park SM Lee HJ Infection following percutaneous vertebral augmentation with polymethylmethacrylate Arch Osteoporos 2018 13 47 29704173 10.1007/s11657-018-0468-y
18 Ma T Lu K Song L Modifiable factors as current smoking, hypoalbumin, and elevated fasting blood glucose level increased the SSI risk following elderly hip fracture surgery J Invest Surg 2020 33 750 8 30885013 10.1080/08941939.2018.1556364
19 Qi H Qi J Sun Y Bone microarchitecture and metabolism in elderly male patients with signs of intravertebral cleft on MRI Eur Radiol 2022 32 3931 43 34989849 10.1007/s00330-021-08458-9
20 Tomé-Bermejo F Piñera AR Duran-Álvarez C Identification of risk factors for the occurrence of cement leakage during percutaneous vertebroplasty for painful osteoporotic or malignant vertebral fracture Spine (Phila Pa 1976) 2014 39 E693 700 24583722 10.1097/BRS.0000000000000294
21 Zampetakis K Stavrakakis IM Samartzidis K Infection following vertebroplasty or kyphoplasty: a systematic review of risk factors, treatment options and outcomes Global Spine J 2026 16 1598 607 41024402 10.1177/21925682251384614 PMC12479454
22 Eckart A Struja T Kutz A Relationship of nutritional status, inflammation, and serum albumin levels during acute illness: a prospective study Am J Med 2020 133 713 22.e7 31751531 10.1016/j.amjmed.2019.10.031
23 Yamamoto Y Shigematsu H Iwata E Hypoalbuminemia increased the length of stay in the treatment of postoperative acute surgical site infection in spinal surgery Spine (Phila Pa 1976) 2020 45 E1564 71 32925680 10.1097/BRS.0000000000003684
24 Zhang F Liu X Tan Z Effect of postoperative hypoalbuminemia and supplement of human serum albumin on the development of surgical site infection following spinal fusion surgery: a retrospective study Eur Spine J 2020 29 1483 9 32048052 10.1007/s00586-020-06306-w
25 Hu W Qiu G Pei F Expert consensus on preoperative nutrition management in major orthopaedic surgery in enhanced recovery after surgery program Chin J Bone Joint Surg 2022 15 763 7
26 Wiedermann CJ Hypoalbuminemia as surrogate and culprit of infections Int J Mol Sci 2021 22 4496 33925831 10.3390/ijms22094496 PMC8123513
27 Weir WR Muraleedharan MV Tuberculosis arising at the site of physical injury: eight case histories J Infect 1983 7 63 6 10.1016/s0163-4453(83)91049-6 6415177
28 Sato K Yamada K Yokosuka K Pyogenic spondylitis: clinical features, diagnosis and treatment Kurume Med J 2019 65 83 9 31406038 10.2739/kurumemedj.MS653001
29 Zheng YH Cao P Chen Z Spinal infection Int J Orthop 2014 35 24 6
30 Zhu J Yang S Cai K Bioactive poly (methyl methacrylate) bone cement for the treatment of osteoporotic vertebral compression fractures Theranostics 2020 10 6544 60 32483469 10.7150/thno.44428 PMC7255031
31 Sugino A Miyazaki T Kawachi G Relationship between apatite-forming ability and mechanical properties of bioactive PMMA-based bone cement modified with calcium salts and alkoxysilane J Mater Sci Mater Med 2008 19 1399 405 17914619 10.1007/s10856-007-3257-5
32 Wendling D Runge M Toussirot E Vertebral osteitis adjacent to kyphoplasty Joint Bone Spine 2010 77 67 9 20031466 10.1016/j.jbspin.2009.11.004
33 Cheung WY Luk KD Pyogenic spondylitis Int Orthop 2012 36 397 404 22033610 10.1007/s00264-011-1384-6 PMC3282872
34 Kim DY Lee SH Jang JS Intravertebral vacuum phenomenon in osteoporotic compression fracture: report of 67 cases with quantitative evaluation of intravertebral instability J Neurosurg 2004 100 24 31 10.3171/spi.2004.100.1.0024 14748570
35 Resnick D Niwayama G Guerra J Spinal vacuum phenomena: anatomical study and review Radiology 1981 139 341 8 7220878 10.1148/radiology.139.2.7220878
36 Dong L Dong C Zhu Y Intravertebral cleft in pathological vertebral fracture resulting from spinal tuberculosis: a case report and literature review BMC Musculoskelet Disord 2020 21 619 32948151 10.1186/s12891-020-03642-2 PMC7501658