Development and validation of a CT-based predictive model for new vertebral compression fractures: the role of vertebral CT attenuation and paraspinal muscle
The nomogram developed based on conventional clinical data in this study, has undergone internal validation and demonstrated efficacy in predicting NVCF following PVP, and serves as a valuable decision-making tool for clinicians.
Abstract
Objectives Osteoporotic vertebral compression fractures (OVCFs) are a major health burden, especially with aging populations. While percutaneous vertebroplasty (PVP) is effective, new vertebral compression fractures (NVCFs) are a common complication. Current NVCF prediction models focus on cement factors or bone density alone, ignoring the critical role of paraspinal muscles. To address this gap, our study integrates preoperative CT imaging of bone density and muscle area with clinical data to develop a visualized early prediction model for NVCF risk, aiming to improve patient outcomes. Methods A total of 423 patients who underwent PVP at Guangzhou University of Chinese Medicine Dongguan hospital were retrospectively analyzed and were allocated into a training set and a validation set in an 8:2 ratio. The study employed multivariate logistic regression analysis and the Least Absolute Shrinkage and Selection Operator (Lasso) to develop predictive models and generate nomogram. Receiver Operating Characteristic (ROC) curves and calibration curves were plotted to evaluate the models’ discriminatory and calibration capabilities, Decision Curve Analysis (DCA) and Clinical Impact Curve(CIC)were used to assess the models’ clinical applicability and utility. Results Lasso regression analysis identified age, albumin (ALB), paravertebral muscle area, bone CT value, low-energy trauma, and single-segment fracture as significant predictors. AUC of the nomogram model in the training set was 0.893 (95% CI: 0.858–0.928) and in the validation set was 0.952 (95% CI: 0.903–1), demonstrating its robust predictive capability. DCA and CIC further suggest that this nomogram model possesses substantial clinical application value. Conclusion The nomogram developed based on conventional clinical data in this study, has undergone internal validation and demonstrated efficacy in predicting NVCF following PVP. It serves as a valuable decision-making tool for clinicians. However, further multicenter studies are required for external validation to confirm its generalizability.
OBJECTIVE
This study aimed to compare the efficacy of endplate Hounsfield unit (HU) values and endplate bone quality (EBQ) scores in predicting cage subsidence (CS) after posterior lumbar interbody fusion (PLIF) in older patients and identify the most discriminative bone mineral density (BMD) assessment indicator.
METHODS
This retrospective analysis included consecutive patients who underwent PLIF at the authors' institution between January 2016 and February 2024. Clinical data were collected for all patients. Propensity scores were used to match patients with and without CS, and the matched cohort was subjected to conditional logistic regression to investigate the association between radiographic factors and CS. L1 and endplate HU values were derived from CT scans, whereas vertebral bone quality (VBQ) and EBQ scores were derived from MR images. Receiver operating characteristic curve analysis was conducted to assess the predictive value of endplate HU values and EBQ scores for CS and further compare their predictive value with that of L1 HU values and VBQ scores.
RESULTS
This study included 130 matched patients. The CS group demonstrated lower L1 (p < 0.001) and endplate HU (p < 0.001) values and higher VBQ (p = 0.002) and EBQ (p < 0.001) scores compared with the non-CS group. The conditional logistic regression analysis identified L1 HU value (OR 0.99, 95% CI 0.97-0.99; p = 0.036), endplate HU value (OR 0.99, 95% CI 0.98-0.99; p = 0.044), VBQ score (OR 2.40, 95% CI 1.34-4.32; p = 0.038), and EBQ score (OR 4.46, 95% CI 2.16-9.18; p = 0.003) as independent predictors of CS, demonstrating areas under the curve of 0.722, 0.815, 0.648, and 0.782, respectively. The optimal cutoff for the endplate HU value in predicting CS was 262.11 (sensitivity 83.08%, specificity 73.85%).
CONCLUSIONS
Endplate HU values indicated a relatively higher predictive performance for CS compared with EBQ scores and served as the most discriminative BMD indicator in patients who underwent PLIF. Measuring the endplate HU value preoperatively helps surgeons select a more appropriate surgical plan and is expected to improve patient outcomes.
Peng Du, Minghui Liang, Ruiyuan Chen et al.· Journal of Neurosurgery : Sp...· 0 citations
Objective To explore the multiple risk factors for intercostal neuralgia after osteoporotic thoracic vertebral compression fractures (OVCF), and to construct a clinical prediction model. Methods The clinical data of 280 patients with single-segment thoracic OVCF admitted to our orthopedic department from January 2022 to December 2025 were retrospectively collected. Patients were categorized into neuropathic pain (NP) and non-neuropathic pain (non-NP) groups based on the Leeds Assessment of Neuropathic Symptoms and Signs (LANSS) scale, with a score of ≥12 defining the primary outcome of intercostal neuralgia as neuropathic pain. Demographic data, fracture-related parameters, bone density, bone metabolism markers, and psychological status (Hospital Anxiety and Depression Scale, HADS) were collected. Univariate and multivariate Logistic regression analyses were used to screen independent risk factors, and the discrimination of the prediction model (area under the ROC curve, AUC) was evaluated. The model’s performance was internally validated using the Bootstrap method. Results Among the 280 patients, the incidence of neuropathic pain was 21.8% (61/280). Multivariate analysis identified six independent risk factors: middle thoracic fracture (T5-T8, OR=4.603), thoracolumbar fascia injury (TLFI, OR=4.883), injured vertebral width ratio (per 0.1 increase, OR=3.973), decreased bone mineral density T-score (per 1-unit decrease, OR=2.685), intravertebral vacuum cleft (IVC, OR=2.764), and depressive state (HADS≥8, OR=2.586). The prediction model showed good calibration (Hosmer-Lemeshow P=0.412) and discrimination (AUC=0.843, 95% CI: 0.789–0.897), with sensitivity 80.3%, specificity 76.7%, and negative predictive value 93.2%. Bootstrap internal validation yielded an optimism-corrected AUC of 0.831. Conclusion Post-fracture intercostal neuralgia in osteoporotic thoracic vertebrae is multifactorial. Middle thoracic vertebra fractures, injury of the thoracolumbar fascia, increased ratio of injured vertebra width, decreased bone density, intravertebral vacuum fissure, and depressive state are independent risk factors. However, external validation in prospective multicenter studies is required before routine clinical implementation.
Dongliang Xiao, Yong-Guang Xu· International Journal of Gen...· 0 citations
Age, number of operated vertebrae, cement leakage, PINP, 25(OH)D, and BMD are closely associated with and are independent risk factors for secondary fractures after PVP/PKP in OTF patients, and the nomogram model based on these factors has high value.
Jiawei Fu, Guanhua Xu, Jia-jia Chen et al.· Frontiers in Surgery· 0 citations
This study aimed to develop and validate a multimodal prediction model integrating biomechanical and radiological variables to predict internal fixation failure in osteoporotic hip fractures, enabling individualized risk assessment and perioperative decision-making. Patients with osteoporotic hip fractures undergoing internal fixation between March 2019 and February 2024 were retrospectively enrolled and randomly divided into a training set (n = 249) and a validation set (n = 107) at a 7:3 ratio. The primary outcome was implant-related failure within 12 months post-surgery. In the training set, univariate analysis and multivariate logistic regression were performed to screen associated factors. Using independent predictors, 3 machine learning models (random forest, support vector machine, and K-nearest neighbors) were developed and compared. The model with the best discriminative ability, assessed by the area under the receiver operating characteristic curve (AUC) with internal validation (bootstrapping), calibration curves, and decision curve analysis, was selected to construct a nomogram. No significant differences in baseline characteristics were observed between the training and validation sets (P > .05). Multivariate logistic regression identified that bone mineral density, maximum fracture end displacement, peak stress distribution of the implant, fracture reduction alignment deviation, and implant insertion depth were significantly associated with fixation failure (P < .05). The nomogram demonstrated excellent performance in both the training (AUC = 0.887, 95% confidence interval: 0.835–0.939) and validation sets (AUC = 0.869, 95% confidence interval: 0.801–0.937). Calibration curves showed good agreement between predicted and observed risks (P > .05), and decision curve analysis indicated superior clinical net benefit across a wide threshold range. Stability testing confirmed no significant multicollinearity (variance inflation factor < 2, events per variable ≥ 5). The random forest model demonstrated strong discriminatory ability and calibration in predicting fixation failure in this single-center retrospective cohort. While the model shows promise for perioperative risk stratification, external validation in multicenter prospective studies is required before clinical implementation.
Background/Objectives: This study aimed to evaluate whether artificial intelligence-derived vertebral volumetric bone mineral density (AI-vBMD) and paraspinal intermuscular adipose tissue (IMAT) ratio from routine computed tomography (CT) could identify moderate-to-severe vertebral compression fractures (VCFs) in breast cancer survivors, and whether paraspinal IMAT ratio and routinely available clinical variables improved diagnostic performance. Methods: This retrospective study included 275 women with breast cancer who underwent routine non-contrast CT and lumbar quantitative computed tomography (QCT). Hounsfield unit-derived volumetric bone mineral density (HU-vBMD) was derived using a QCT-referenced HU-to-vBMD conversion equation, whereas AI-vBMD and paraspinal IMAT ratio were extracted using automated software. Moderate-to-severe VCF was defined as Genant grade ≥ 2. Agreement with QCT-vBMD was assessed using correlation, intraclass correlation coefficient (ICC), and Bland–Altman analysis. Model discrimination was evaluated using receiver operating characteristic analysis and DeLong tests. Results: Moderate-to-severe VCF was present in 75 patients (27.3%). HU-vBMD and AI-vBMD showed excellent agreement with QCT-vBMD (ICC, 0.978 and 0.987, respectively). AI-vBMD outperformed HU-vBMD for identifying VCFs (AUC, 0.738 vs. 0.714; p < 0.001). IMAT ratio showed comparable standalone discrimination to AI-vBMD (AUC, 0.760 vs. 0.738; p = 0.604). Adding IMAT ratio to AI-vBMD improved discrimination (AUC, 0.786 vs. 0.738; p = 0.038). The full model incorporating clinical covariates achieved the highest AUC (0.828; 95% CI, 0.778–0.878). Conclusions: AI-vBMD and paraspinal IMAT ratio automatically extracted from routine CT improved the diagnostic assessment of prevalent moderate-to-severe VCFs in breast cancer survivors. This study supports an automated CT-based approach that integrates vertebral bone density and paraspinal muscle–fat information for opportunistic identification of clinically relevant VCFs.
Chen Wan, Lingquan Kong, Jie Hao et al.· Journal of Clinical Medicine· 0 citations
Background and Objectives: Osteoporotic vertebral compression fractures affect approximately one in four postmenopausal women and carry substantial morbidity, yet established clinical tools such as dual-energy X-ray absorptiometry (DXA) provide only patient-level risk and do not identify which specific vertebra is most likely to fail. Computed tomography (CT) acquired for unrelated indications is the most widely available three-dimensional substrate for opportunistic screening, but published machine learning models for vertebral fracture risk almost universally operate at the patient level. The present study aimed to develop and rigorously validate a per-vertebra prediction pipeline applicable to both routine clinical lumbar-spine CT and opportunistic abdominal CT, both acquired for indications unrelated to osteoporosis screening. Materials and Methods: Two independent retrospective cohorts were assembled from a single academic centre: a routine clinical lumbar-spine CT cohort of 106 patients yielding 478 evaluable vertebrae, and a routine abdominal CT cohort of 126 patients yielding 589 evaluable vertebrae. Vertebral bodies were segmented automatically with TotalSegmentator v2 and the trabecular core isolated by morphological erosion. A panel of 505 quantitative imaging biomarkers compliant with Image Biomarker Standardisation Initiative recommendations was extracted, covering trabecular density, vertebral morphometry, classical texture, trabecular network architecture, sub-endplate vulnerability, low-density topology, radial heterogeneity and adjacent muscle quality. Within-patient feature engineering expanded the input pool to 1293 contextual descriptors. Three model families were evaluated under fully nested leave-one-patient-out cross-validation: ElasticNet logistic regression, a softmax-ranking approximation of conditional logistic regression, and a Two-Stage model combining a patient-level fragility score with a within-patient vertebral outlier score. Patient-level bootstrap resampling (2000 iterations) was used to obtain 95% confidence intervals. Results: On routine clinical lumbar-spine CT the Two-Stage model achieved a per-vertebra AUC of 0.750 (95% CI 0.704 to 0.795), an F1 of 0.549, a within-patient concordance index of 0.693, an expected calibration error of 0.044, and Hit@3 of 0.934. It was the only model evaluated that returned calibrated probabilities; the softmax-ranking and ElasticNet baselines gave expected calibration errors of 0.232 and 0.218 respectively. On opportunistic abdominal CT, the softmax-ranking model gave AUC 0.672 (95% CI 0.615 to 0.727). Selected biomarkers were dominated by regional trabecular density and trabecular network architecture; a stable core of lumbar features entered the model in 100% of cross-validation folds, indicating high reproducibility. The closest prior per-vertebra CT-based predictor in primary, non-surgical patients (Muehlematter and colleagues, 58-patient cohort) reported a per-vertebra AUC of 0.64, which is one of several reference points for the present results. Ten methodological variants and sensitivity analyses, including rank fusion, internal tissue normalisation and additional biomechanical features, did not provide statistically significant gains, indicating that the binding constraint at this sample size is data volume rather than methodology. Conclusions: A two-stage decomposition that separates systemic skeletal fragility from within-patient vertebral outlier status produces well-calibrated per-vertebra fracture-risk estimates from routine clinical lumbar spine CT and was the only model evaluated to do so, which is what permits a per-vertebra output to be reported as an absolute risk rather than as an ordering alone; a within-patient ranking model is preferable for opportunistic abdominal CT. The discrimination advantage of the decomposition over that baseline is numerical and consistent but not statistically established at this sample size, and the work is presented as a transparent and reproducible single-centre benchmark for the still under-developed per-vertebra prediction task. Its clearest near-term value is opportunistic, namely flagging elevated per-vertebra fracture risk on CTs already acquired for unrelated indications without additional radiation, cost or a dedicated densitometric study. External multi-centre validation is the necessary next step.
K. Riazanovskiy, Dāvids Orlovs, Jekaterina Stepanova et al.· Medicina· 0 citations
