Predictive Value of the Preoperative Screw Trajectory-to-Vertebral Body Hounsfield Unit Ratio and the Combined Model Incorporating MRI-Based Pedicle Bone Quality Score for Pedicle Screw Loosening After Pedicle Screw Fixation

Abstract

Study Design

Retrospective cohort study.

Objectives

This study introduced the screw trajectory-to-vertebral body Hounsfield unit (HU) ratio as a novel parameter, evaluated its predictive value with the pedicle bone quality (PBQ) score for postoperative pedicle screw loosening, and develop a nomogram for risk prediction.

Methods

Retrospective analysis included 313 patients undergoing pedicle screw fixation (2022-2024). Screw loosening was defined as a ≥1.0-mm radiolucent zone or double-halo sign on postoperative computed tomography (CT) or radiographs. Preoperative CT and magnetic resonance imaging (MRI) were used to calculate the HU ratio and PBQ score. Candidate predictors were selected based on univariable screening and clinical relevance; multivariable logistic regression was used to develop the prediction model and nomogram. Model performance was evaluated by area under the curve (AUC), Hosmer–Lemeshow test, and bootstrap validation. A sensitivity analysis was performed by additionally including 43 scoliosis patients (n = 356).

Results

At 12 months follow-up, the loosening rate was 14.3% (45/313) in the patients. Independent risk factors included higher PBQ score, lower HU ratio, older age, and higher Goutallier grade. The nomogram showed excellent discrimination (apparent AUC = 0.904, 95% confidence interval [CI]: 0.856-0.953; optimism-corrected AUC = 0.883) and acceptable calibration (Hosmer–Lemeshow P = .393). At the optimal cutoff (predicted probability ≥0.212), sensitivity was 80.0% and specificity 88.1%. In sensitivity analysis, construct-related factors (fusion length and terminal fusion segment) became statistically significant, while overall model performance remained comparable.

Conclusions

The HU ratio is an independent predictor of postoperative pedicle screw loosening and complements the PBQ score.

Keywords

hounsfield unit ratio pedicle bone quality score computed tomography magnetic resonance imaging pedicle screw loosening nomogram

Introduction

Pedicle screw fixation is a standard surgical intervention in spine surgery for degenerative spinal diseases and scoliosis, playing a pivotal role in maintaining spinal stability and promoting bony fusion.¹ However, postoperative pedicle screw loosening is a common clinical complication that may result in screw breakage, suboptimal fusion or pseudarthrosis, chronic pain, progression of kyphosis, and other adverse outcomes.² Previous studies have demonstrated that the rate of postoperative pedicle screw loosening is significantly higher in patients with osteoporosis compared to non-osteoporotic individuals.^3,4 Moreover, osteoporotic bone exhibits two characteristic features: fatty infiltration and trabecular atrophy.⁵ Consequently, accurate preoperative assessment of bone quality and status is essential for predicting and mitigating the risk of pedicle screw loosening. Historically, dual-energy X-ray absorptiometry (DXA) has been the clinical gold standard for evaluating bone quality.^6,7 However, degenerative changes in the lumbar spine (eg, osteophytes and facet hypertrophy) can artifactually elevate DXA measurements, leading to an overestimation of bone mineral density (BMD).⁸ Lower lumbar segments, such as L3-L4, are particularly susceptible to unreliable T-scores; patients who are older, have a higher BMI, or have concomitant degenerative scoliosis are more likely to exhibit a greater number of segments with unreliable T-scores.⁹ As a result, recent years have seen the emergence of several imaging-based alternative assessment metrics, including computed tomography (CT)-based Hounsfield unit (HU) measurements and magnetic resonance imaging (MRI)-based bone quality scoring systems, such as the pedicle bone quality (PBQ) score.⁷

In 2011, Schreiber JJ proposed assessing bone quality by measuring vertebral body or pedicle HU values on routine lumbar computed tomography, finding that HU values were significantly associated with postoperative pedicle screw loosening after fixation.¹⁰ Compared to DXA, HU better reflects local mineralization, but its absolute value is highly influenced by an individual’s overall bone mass, scan parameters, and tissue architecture.¹¹ Clinically, two scenarios are difficult to distinguish by “absolute HU”: (1) globally reduced bone mass with a relatively preserved pedicle screw trajectory, and (2) overall bone status that is acceptable but with focal weakness along the pedicle screw trajectory. Moreover, even when overall bone quality is suboptimal, adequate mineralization of cortical and cancellous bone along the screw trajectory may still provide sufficient fixation strength,¹² reducing the risk of loosening. Prior studies have also suggested that measuring HU along the planned screw trajectory is more representative.¹³ Given the significant interpatient differences in overall bone mass, simple comparisons of local absolute HU lack comparability and are easily influenced by individual variability.¹⁴ Based on these considerations, the concept of the HU ratio was introduced: the ratio of HU measured along the planned screw trajectory to HU within the vertebral body at the same level. This ratio quantifies the relative bone quality of the anchoring region in relation to the overall load-bearing bone, providing a unified, comparable scale amidst interindividual variability. Screw stability is influenced by multiple factors, with fatty infiltration of local cancellous bone and degeneration of trabecular microarchitecture particularly affecting anchorage strength. An increase in bone marrow adipocytes is closely associated with remodeling of the trabecular network and is a key phenotype of osteoporosis.⁴ Under osteoporotic conditions, increased bone marrow adipocytes impair compensatory remodeling of the trabecular microarchitecture.^5,7 Osteoporosis is commonly characterized by marrow adipocyte infiltration and trabecular atrophy.¹⁵ However, the HU ratio primarily reflects mineralization distribution and has limited capacity to indicate the degree of fatty infiltration, highlighting the need for complementary indicators.

In 2024, building on vertebral bone quality (VBQ), researchers proposed an MRI-based PBQ score, defined as the ratio of the mean signal intensity in the pedicle regions from L1-L4 to the cerebrospinal fluid signal intensity at the L3 level.¹⁶ This score is designed to reflect the degree of marrow fatty infiltration in the screw insertion region, and subsequent studies have demonstrated its value in predicting pedicle screw loosening and assessing marrow fatty infiltration.^13,16 However, the PBQ score has limited sensitivity to cortical bone thickness and mineralization.^7,13 Conceptually, the PBQ score primarily reflects marrow fatty infiltration and degeneration of the trabecular microarchitecture in the pedicle region, while the HU ratio emphasizes the relative distribution pattern of mineralization. Accordingly, the hypothesis that the two measures are complementary was proposed and tested in this study.

This study introduced and defined the HU ratio as a CT-based parameter to be used alongside the MRI-based PBQ score and evaluated additional spinal structural factors (eg, terminal fusion segment) in assessing the risk of postoperative pedicle screw loosening. The independent predictive effects of the PBQ score and the HU ratio were examined, and their complementarity in capturing marrow fatty infiltration and mineralization distribution was evaluated. Furthermore, both measures, along with other clinically relevant factors, were incorporated into the development and validation of a nomogram for predicting pedicle screw loosening, with the goal of improving the accuracy of preoperative risk assessment.

Methods

After ethical approval, the study retrospectively reviewed the clinical data of patients who underwent pedicle screw fixation in the Department of Spine Surgery from January 2022 to September 2024. Inclusion criteria were: (1) age ≥18 years; (2) pedicle screw fixation performed at our institution with follow-up ≥ 12 months for patients without adult degenerative scoliosis (ADS) and ≥24 months for ADS patients; (3) completion of preoperative lumbar MRI and CT; (4) postoperative radiographs or CT indicating the presence or absence of screw loosening; and (5) MRI and CT findings correlated with clinical symptoms. Exclusion criteria were: (1) any history of lumbar spine surgery (including prior lumbar fusion, instrumentation, or implant removal); (2) non-degenerative lumbar conditions such as diffuse idiopathic skeletal hyperostosis, ankylosing spondylitis, lumbar infection, lumbar tumor, or lumbar trauma; (3) poor image quality or metal artifacts; (4) incomplete surgical information or incomplete preoperative/postoperative imaging; (5) severe medical comorbidities or underlying metabolic bone disease; and (6) postoperative spinal infection (superficial or deep) identified during follow-up.

Data Collection and Imaging Assessment

Demographic and clinical data were collected and recorded, including age, sex, body mass index (BMI), diabetes mellitus, number of fused levels, surgical technique, and terminal fusion segment. All operations were performed by the same surgical team following an institutional standardized fixation strategy (including implant selection, screw type/configuration, and trajectory planning whenever feasible) to minimize technique-related variability, and the same general anesthesia protocol was applied to all patients. Two experienced spine surgeons independently assessed pedicle screw loosening and bony fusion status, with final determinations reached by consensus after discussion. All imaging parameters were quantified by two experienced surgeons who were blinded to each other’s assessments. Screw loosening was assessed using postoperative radiographs or CT obtained at the last follow-up visit (≥12 months after surgery; ≥24 months for patients with adult degenerative scoliosis). Screw loosening was defined as a radiolucent zone ≥1.0 mm in width around any implanted screw or the presence of a “double-halo” sign on postoperative CT or radiographs.¹⁷

(1) Measurement of vertebral body and screw-trajectory HU values

All patients underwent preoperative lumbar CT (Definition, Siemens). The tube voltage of the CT scans was set at 120 kV. Evaluation of CT scans and region of interest (ROI)-based quantification were performed on the institutional picture archiving and communication system (INFINITT PACS, version 3.0.11.4-BN11; INFINITT Healthcare, Seoul, Republic of Korea), which was used to compute the mean HU for each ROI.¹⁸ On axial sections of L3, the largest possible elliptical ROI was drawn within the cancellous bone to measure HU; values were obtained on three planes—just below the superior endplate, mid-vertebral, and just above the inferior endplate—and the mean of the three measurements was used to represent the vertebral body trabecular HU (B-HU).^19,20 Screw-trajectory HU (ST-HU) was obtained by drawing a ROI on axial preoperative CT images at the level and orientation of each pedicle screw. Postoperative CT scans were used as a visual reference to identify the final position and angulation of each screw, and corresponding ROIs were then manually placed on the preoperative images so that their location and direction approximated the actual screw trajectory, thereby reflecting the bone density along the intended anchorage corridor. On axial sections at L3, the left- and right-side trajectory HU measurements were averaged to represent the ST-HU for that level.^12,21 The HU ratio was calculated as ST-HU/B-HU (ST-HU divided by B-HU) and used as an index of local screw-trajectory bone quality relative to the overall vertebral body. Given that postoperative screw loosening results from multiple contributing factors, this study prespecified L3 as the primary measurement level (substituting L2 or L4 when L3 was not measurable due to focal pathology or artifacts) to minimize potential influences beyond the PBQ score and the HU ratio.²⁰ The measurement method of B-HU and ST-HU is illustrated in Figure 1.

(2) Measurement of PBQ

Figure 1.

Measurement of vertebral body HU (B-HU) and screw-trajectory HU (ST-HU). (A-B): rectangular regions of interest (ROIs) were placed along the intended pedicle screw trajectory; (C-E): circular ROIs were positioned in the cancellous bone area of the vertebra, corresponding to the three levels localized on the sagittal plane

Among the included patients, lumbar MRI examinations were performed using a Philips Achieva 3.0 T scanner. Preoperative sagittal T1-weighted images of the lumbar spine were used to measure the PBQ score as follows: For the pedicle channels from L1 to L4, ROIs were placed approximately 1 mm from the cortical margin to measure the signal intensity (SI) of the pedicular cancellous bone. The mean SI of the bilateral pedicles was recorded as the segmental SI. An additional ROI was placed in the subarachnoid space posterior to the L3 vertebral body, avoiding nerve roots, to measure the cerebrospinal fluid (CSF) SI. If CSF was affected at the L3 level, CSF measurements at the L2 or L4 level were selected. The PBQ score was then calculated by dividing the mean SI of the L1-L4 pedicles by the SI of the CSF at the L3 level.^7,22 A representative example of PBQ score measurement is shown in Figure 2.

(3) Dual-energy X-ray absorptiometry (DXA)

Figure 2.

The demonstration of the PBQ score measurement in a representative non-contrast MRI T1-weighted sagittal lumbar image. (A): Placement of the region of interest (ROI) in the pedicle channels from the L1 to L4 segments; (B): Placement of the ROI in the subarachnoid space posterior to the L3 vertebral body

DXA of the lumbar spine and proximal femora was performed using standard techniques on Lunar Prodigy densitometers (GE Healthcare, Waukesha, Wisconsin). As recommended by the World Health Organization, osteoporosis was defined as a T-score below −2.5 SD of the BMD, while osteopenia was defined as a T-score between −1.0 and −2.5 SD.^23,24 At least one valid reported T-score for the lumbar spine or hips was required for study inclusion.

(4) Qualitative Assessment of Paraspinal Muscle Fatty Infiltration

A qualitative assessment of paraspinal muscle fatty infiltration (Goutallier grade) was performed using the Goutallier classification system. The muscle composition of the multifidus muscle on MRI was classified independently by two readers into five different grades based on the visually assessed fat/muscle ratio at each disc level from L1/L2 to L5/S1 (five segments in total). The severity of fatty infiltration was graded as follows: grade 0 (no fatty streaks), grade 1 (some fatty streaks), grade 2 (fatty infiltration but still more muscle fibers than fat), grade 3 (equal amounts of fat and muscle fibers), and grade 4 (larger amounts of fat than muscle fibers). The higher Goutallier grade between those of the right and left multifidus was selected and used in the analysis.

(5) Measurement of spinopelvic parameters

Standing long-cassette lateral radiographs were obtained with the patient in a comfortable, natural upright posture. Spinopelvic parameters were treated as secondary covariates and measured using standard definitions: pelvic incidence (PI)—the angle between the line from the S1 endplate center to the bicoxofemoral axis and the perpendicular to the S1 endplate; pelvic tilt (PT)—the angle between the vertical and the line from the S1 endplate center to the bicoxofemoral axis; and sacral slope (SS)—the angle between the S1 endplate and the horizontal. By definition, PI = PT + SS. Lumbar lordosis (LL) was measured as the Cobb angle between the superior endplate of L1 and the superior endplate of S1.

Statistical Analysis

Statistical computations were performed using the Statistical Package for the Social Sciences (SPSS Inc., Chicago, IL, USA). For cases with missing data, multiple imputation was employed, utilizing chained equations (FCS; m = 5, 10 iterations per chain) and pooled estimates via Rubin’s rules. The variables with missing data and their corresponding missing proportions are summarized in Supplemental Table 6. Normally distributed continuous variables are presented as mean ± standard deviation, while non-normally distributed data are presented as median (interquartile range). Normality was assessed using the Shapiro–Wilk test, with between-group comparisons conducted using the t-test or the Mann–Whitney U test, and categorical variables analyzed with the chi-square test. The correlation between the PBQ score and the L3 pedicle-to-vertebral body HU ratio was evaluated using Pearson and Spearman correlation coefficients, followed by collinearity diagnostics. Multicollinearity among predictors was assessed using tolerance, variance inflation factors (VIFs), and the condition index. Screw loosening served as the dependent variable. Variables with P < .05 in univariate logistic regression and variables considered clinically important a priori (age, number of fused levels, PBQ score, HU ratio, Goutallier grade, and terminal fusion segment) were entered into the multivariable logistic regression model, with results expressed as odds ratios (ORs) and 95% confidence intervals (CIs). Discrimination was assessed using receiver operating characteristic (ROC) curves and the area under the curve (AUC) for each predictor retained in the final model (age, PBQ score, HU ratio, and Goutallier grade) as well as for the combined multivariable model (predicted probability). Optimal cutoff values, sensitivity, specificity, and the Youden index were derived using MedCalc (version 23), with the maximum Youden index calculated as sensitivity + specificity – 1.²⁵ Because missing high-risk patients could lead to screw loosening, re-operation and long-term pain, whereas false positives only lead to preventative measures such as cement augmentation with relatively low cost, we prioritized higher sensitivity. Model calibration was evaluated using the Hosmer–Lemeshow goodness-of-fit test and logistic recalibration, summarized by the calibration intercept (calibration-in-the-large, CITL) and slope with 95% CIs. Internal validation was performed through bootstrap resampling with 1000 repetitions. In each bootstrap sample, the model was refitted, and its performance was evaluated both in the bootstrap sample and the original dataset; the average optimism was then subtracted from the apparent estimates to obtain optimism-corrected AUC, Brier score, CITL, and calibration slope. As a sensitivity analysis, we repeated the analyses in an expanded cohort additionally including 43 patients with adult degenerative scoliosis (ADS) to assess robustness across a broader clinical spectrum; baseline characteristics, regression results, and model performance are summarized in Supplemental Tables S1-S3. Given the conceptual coupling between construct-related variables (eg, LIV and fusion length), we further examined their association and collinearity in the expanded cohort (Supplemental Tables S4-S5).

Results

A total of 487 patients were screened, with 131 excluded based on prespecified criteria: 56 due to poor image quality or metal artifacts, 24 for missing preoperative MRI or CT, and 51 for loss to follow-up or missing outcome data. Ultimately, a total of 356 patients were eligible for analysis. To reduce etiological heterogeneity, we predefined the primary analysis cohort by excluding patients with ADS, and finally obtained 313 patients for model construction and primary inference, of whom 45 (14.3%) experienced pedicle screw loosening. The mean age was 62.620 ± 10.162 years; 118 (37.7%) were women and 195 (62.3%) were men. The mean BMI was 25.054 ± 2.961 kg/m² (Table 1). Furthermore, an expanded cohort of patients with ADS (n = 356, including 43 ADS cases) was incorporated into the analysis as a sensitivity analysis. This was conducted to verify the robustness of our primary findings across a broader clinical spectrum and to evaluate the impact of surgery-related factors with greater variability. Baseline characteristics of the expanded cohort are shown in Supplemental Table 1. Univariable and multivariable analyses in the expanded cohort are provided in Supplemental Table 2, and model validation is reported in Supplemental Table 3.

Table 1.

Baseline Characteristics of the Included Patients

Variable	Overall (n = 313)	Loosening (n = 45)	Non-loosening (n = 268)	P-value
Demographic and clinical variables
Age (years), mean ± SD	62.620 ± 10.162	68.193 ± 9.914	61.684 ± 10.202	<.001
Sex, n (%)				.499
Female	118 (37.7)	19 (42.2)	99 (36.9)
Male	195 (62.3)	26 (57.8)	169 (63.1)
BMI (kg/m²), mean ± SD	25.054 ± 2.961	24.971 ± 3.374	25.068 ± 2.887	.840
Diabetes mellitus, n (%)				.236
No	243 (77.6)	38 (84.4)	205 (76.5)
Yes	70 (22.4)	7 (15.6)	63 (23.5)
Length of stay (days), mean ± SD	8.348 ± 2.301	8.244 ± 2.451	8.366 ± 2.275	.744
Bone mineral density (DXA), mean ± SD (T-score)	−0.726 ± 1.265	−0.773 ± 1.389	−0.719 ± 1.244	.790
Spinopelvic and sagittal parameters, mean ± SD
PI (°)	55.452 ± 11.276	58.045 ± 12.737	55.017 ± 11.017	.097
PT (°)	28.649 ± 8.866	28.934 ± 7.998	28.602 ± 9.001	.816
SS (°)	26.818 ± 10.731	29.111 ± 11.098	26.434 ± 10.669	.122
Lumbar lordosis (LL, °)	39.989 ± 13.206	34.019 ± 13.893	40.992 ± 13.089	.001
Surgical variables
Number of fused levels, mean ± SD	2.578 ± 0.683	3.000 ± 0.853	2.508 ± 0.651	<.001
Surgical type, n (%)				.049
A: short lumbar fusion	207 (66.1)	23 (51.1)	184 (68.7)
B: lumbosacral fusion	97 (31.0)	21 (46.7)	76 (28.4)
C:long lumbar/thoracolumbar fusion	9 (2.9)	1 (2.2)	8 (3.0)
Terminal fusion segment, n (%)				.108
L5	199 (63.6)	23 (51.1)	176 (65.7)
S1	91 (29.1)	19 (42.2)	72 (26.9)
L4 or above	23 (7.3)	3 (6.7)	20 (7.5)
Radiologic muscle and bone quality
Paraspinal fatty infiltration grade, n (%)				<.001
Grade 1	52 (16.6)	4 (8.9)	48 (17.9)
Grade 2	220 (70.3)	25 (55.6)	195 (72.8)
Grade 3	10 (3.2)	5 (11.1)	5 (1.9)
Grade 4	31 (9.9)	11 (24.4)	20 (7.5)
PBQ score	3.084 ± 0.607	3.800 ± 0.651	2.963 ± 0.600	<.001
HU ratio	1.841 ± 0.560	1.502 ± 0.754	1.861 ± 0.522	<.001

Abbreviations: BMI, body mass index; DXA, dual-energy X-ray absorptiometry; PI, pelvic incidence; PT, pelvic tilt; SS, sacral slope; LL, lumbar lordosis; PBQ, pedicle bone quality; HU, Hounsfield unit; SD, standard deviation.

Patients were divided into a loosening group and a non-loosening group based on the presence of pedicle screw loosening. In univariate logistic regression, age, number of fused levels, PBQ score, HU ratio, paraspinal fatty infiltration grade were significantly associated with screw loosening, whereas BMD, length of stay, BMI, pelvic parameters (PI, PT, SS), sex, and diabetes mellitus were not (all P > .05; Table 2). Variables with P < .05 in univariate logistic regression, as well as variables considered clinically important a priori, were entered into the multivariable logistic regression model. Given that surgical type and terminal fusion segment both reflect the distal construct, they were not simultaneously included in the multivariable analysis, only terminal fusion segment was retained in the fully adjusted model, and the LL angle did not remain an independent predictor. After full adjustment, however, number of fused levels and terminal fusion segment were not statistically significant (all P > .05) and were therefore excluded from the final model (Table 2). The loosening group had a significantly higher PBQ score (3.800 ± 0.651 vs 2.963 ± 0.600; P < .001) and a significantly lower HU ratio (1.502 ± 0.754 vs 1.861 ± 0.522; P < .001). Multivariate logistic regression confirmed both indicators as predictors of screw loosening (PBQ: adjusted OR [aOR] = 8.626, 95% CI 4.233-17.575; HU ratio: aOR = 0.187, 95% CI 0.080-0.440; both P < .001). In the same multivariate model, higher PBQ score and lower HU ratio were independently associated with an increased risk of pedicle screw loosening (both P < .001).

Table 2.

Univariate Logistic Regression Analyses and Multivariate Logistic Regression Analyses of Factors Associated With Screw Loosening in Patients Following Pedicle Screw Fixation

Factor	Univariate logistic regression analysis		Multivariate logistic regression analysis
Factor	OR (95% CI)	P-value	aOR (95% CI)	P-value
Age	1.072 (1.034-1.111)	<.001	1.070 (1.023-1.120)	.003
Bone mineral density (DXA), mean ± SD (T-score)	0.967 (0.755-1.239)	.789	—	—
Length of stay (per day)	0.977 (0.851-1.122)	.743	—	—
BMI (kg/m²)	0.989 (0.888-1.101)	.839	—	—
Number of fused levels (per level)	2.425 (1.591-3.695)	<.001	—	—
PI (°)	1.024 (0.996-1.054)	.098	—	—
PT (°)	1.004 (0.969-1.041)	.815	—	—
SS (°)	1.026 (0.993-1.059)	.122	—	—
LL angle (°)	0.964 (0.942-0.986)	.002	—	—
PBQ	7.418 (4.114-13.376)	<.001	8.626 (4.233-17.575)	<.001
HU ratio	0.191 (0.084-0.433)	<.001	0.187 (0.080-0.440)	<.001
Sex	0.802 (0.422-1.522)	.499	—	—
Diabetes mellitus	1.668 (0.710-3.919)	.240	—	—
Goutallier grade	—	—	—	—
Grade 1 vs grade 4	0.152 (0.043-0.533)	.003	0.126 (0.025-0.627)	.011
Grade 2 vs grade 4	0.233 (0.100-0.543)	.001	0.228 (0.067-0.775)	.018
Grade 3 vs grade 4	1.818 (0.430-7.685)	.416	1.895 (0.257-13.999)	.531
Surgical type (ordinal)	1.695 (0.987-2.913)	.056	—	—
Terminal fusion segment	—	—	—	—
L5 vs L4 or above	0.568 (0.153-2.114)	.400	—	—
S1 vs L4 or above	0.495 (0.123-1.996)	.312	—	—

Abbreviations: OR, odds ratio; aOR, adjusted odds ratio; CI, confidence interval; BMI, body mass index; DXA, dual-energy X-ray absorptiometry; PI, pelvic incidence; PT, pelvic tilt; SS, sacral slope; LL, lumbar lordosis; PBQ, pedicle bone quality; HU, Hounsfield unit; SD, standard deviation.

Odds ratios are expressed per unit increase unless otherwise specified. Grade 4 was used as the reference category for Goutallier grade, and L4 or above was used as the reference category for terminal fusion segment.

Footnote: The numbers of patients in each Goutallier grade were as follows: grade 1 (n = 52), grade 2 (n = 220), grade 3 (n = 10), grade 4 (n = 31). Caution should be exercised in interpreting the results for subgroups with small sample sizes.

Accordingly, the final prediction model and nomogram were constructed based on the four variables that remained significant in multivariable analysis: age, PBQ score, HU ratio, and paraspinal muscle fatty infiltration (Goutallier grade) (Table 2). The PBQ score was not substantially correlated with the HU ratio (Spearman ρ = 0.039, P = .491). Collinearity diagnostics indicated no evidence of multicollinearity among the four candidate predictors considered during model development (tolerance = 0.905-0.985; VIF = 1.015-1.105; maximum condition index = 19.854). In the expanded cohort, LIV (terminal segment) was significantly associated with instrumentation length (Supplemental Table S4), supporting that these construct descriptors capture overlapping construct characteristics. Accordingly, we assessed potential collinearity between LIV and instrumentation length (Supplemental Table S5) and avoided simultaneously including highly overlapping construct-related variables in the primary model specification.

Model Construction

Based on the multivariate logistic regression model, a nomogram (Figure 3) was developed incorporating four independent predictors: age, PBQ score, HU ratio, and the Goutallier paraspinal muscle fatty infiltration grade. For each patient, the value of each predictor is located on its corresponding axis and projected upward to the “Points” scale to obtain an individual point score. The total score is calculated by summing the points across all four predictors, and the predicted probability of loosening is then determined by locating the total score on the “Total Points” axis and projecting downward to the “Risk” axis. For example, a 70-year-old patient with a PBQ score of 3.8, an HU ratio of 1.5, and a Goutallier grade of 3 would receive approximately 44, 56, 73, and 44 points, respectively, resulting in a total of about 217 points. This total corresponds to an estimated loosening risk of approximately 95% on the nomogram, which exceeds the optimal probability cutoff of 0.212 (21.2%) identified in this study. This indicates a high-risk patient, for whom augmented fixation or bone-strengthening strategies should be considered.

Figure 3.

Nomogram for predicting the risk of pedicle screw loosening after pedicle screw fixation

Model Validation

ROC analysis was used to evaluate the discriminative ability of the combined prediction model (logistic regression–based predicted probability underlying the nomogram) and each individual predictor. Because the HU ratio was inversely associated with screw loosening, ROC analysis for this variable was performed with a reversed test direction (smaller test result indicating higher risk). The AUCs, optimal thresholds, sensitivities, specificities, and Youden indices for all predictors are summarized in Table 3. The combined multivariable model showed excellent discrimination, with an AUC of 0.904 (95% CI 0.856-0.953), as illustrated by the ROC curve in Figure 4, whereas the individual predictors provided only moderate discriminative ability. Model calibration was assessed using the Hosmer–Lemeshow goodness-of-fit test, calibration plots, and logistic recalibration. The calibration curve (Figure 5) showed good agreement between predicted and observed risks, and the Hosmer–Lemeshow test (χ² = 8.423, df = 8, P = .393) did not indicate a significant lack of fit. Bootstrap internal validation with 1000 resamples showed only modest optimism: the apparent C-index (AUC) of 0.904 decreased to an optimism-corrected value of 0.883, and the Brier score increased from 0.074 to 0.083. After bootstrap correction, the calibration intercept (CITL) was −0.188 (95% CI −0.960 to 0.451) and the calibration slope was 0.817 (95% CI 0.277 to 1.222), suggesting mild overall risk overestimation but preserved discrimination. Decision curve analysis further demonstrated a higher net benefit of using the nomogram than either treat-all or treat-none strategies across a clinically relevant range of threshold probabilities. Model performance remained comparable in the expanded cohort, with consistent discrimination and calibration after bootstrap correction (Supplemental Table S3).

Table 3.

Receiver Operating Characteristic (ROC) Analysis of Screw Loosening in Patients Following Pedicle Screw Fixation

Factor	AUC	95% CI	P-value	Sensitivity	Specificity	Youden index	Optimal threshold
Age	0.676	0.584-0.768	<.001	0.733	0.545	0.278	62.335
Goutallier grade	0.647	0.554-0.740	.002	0.356	0.907	0.263	2.5
PBQ score	0.818	0.746-0.889	<.001	0.778	0.776	0.558	3.437
HU ratio	0.741	0.653-0.829	<.001	0.644	0.743	0.573	1.521
Predicted probability (combined model)	0.904	0.856-0.953	<.001	0.800	0.881	0.681	0.212

Abbreviations: ROC, receiver operating characteristic; AUC, area under the curve; CI, confidence interval; PBQ, pedicle bone quality; HU, Hounsfield unit.

The Youden index was calculated as sensitivity + specificity −1.

The optimal threshold was determined by maximizing the Youden index.

Predicted probability was derived from the multivariable logistic regression model.

Figure 4.

ROC curves for predicting pedicle screw loosening after pedicle screw fixation

Figure 5.

Calibration curve obtained from internal validation using bootstrap resampling (B = 1,000 repetitions; n = 313). Apparent, bias-corrected, and ideal curves are shown, with a mean absolute error (MAE) of 0.054, indicating good agreement between predicted and observed probabilities

Clinical Practice

We performed decision curve analysis (DCA) to evaluate the clinical utility of the nomogram (Figure 6). In the DCA plot, the dashed curve represents the strategy of treating all patients, the solid line represents the strategy of treating no patients, and the red curve represents the net benefit of our model. The results showed that our model provided a superior clinical net benefit within a threshold probability range of approximately 0.05-0.80. Based on the DCA curve, a clinical impact curve was plotted for a hypothetical cohort of 1000 patients undergoing pedicle screw fixation (Figure 7). The solid black line (“Number high risk/1000”) represents the number of patients classified as high risk of screw loosening by the model at each risk threshold, and the dashed red line (“Number high risk with event/1000”) represents the number of these high-risk patients who would actually develop pedicle screw loosening (true positives). Across a wide range of thresholds, the number of true positives was relatively high compared with the total number classified as high risk, indicating that the model has reasonable clinical utility for identifying patients at increased risk of postoperative pedicle screw loosening.

Figure 6.

Decision curve analysis of the nomogram. The red line represents the model. The x-axis and y-axis show the threshold probability and net benefit, respectively. The gray line represents the net benefit of the treat-all strategy. The horizontal black line represents the net benefit of the treat-none strategy (no intervention)

Figure 7.

The horizontal axis shows the risk threshold and the corresponding cost–benefit ratio, and the vertical axis shows the number of patients per 1000 classified as high risk. The solid black curve (“Number high risk /1000”) indicates the number of patients who are classified by the model as having a high risk of postoperative pedicle screw loosening at each threshold probability, whereas the dashed red curve (“Number high risk with event /1000”) indicates the number of these high-risk patients who would actually develop pedicle screw loosening at each threshold probability

Discussion

This study introduced a novel CT-based parameter, the HU ratio, as a preoperative bone quality measure to predict the risk of pedicle screw loosening after pedicle screw fixation. This metric reflects the relative degree of mineralization along the planned screw trajectory compared to the vertebral body, thereby partially mitigating the confounding effect of interindividual differences in bone quality on risk assessment. Building on this, a combined predictive model was developed, incorporating the MRI-based PBQ score (reflecting marrow fatty infiltration) and the HU ratio, thus integrating two complementary aspects of bone quality: mineralization and fatty infiltration. The findings indicate that the HU ratio and PBQ score are independent predictors, with the combined model providing superior discrimination and calibration, substantially improving the preoperative identification of high-risk patients.

To reduce measurement heterogeneity, L3 was used as the primary measurement level; prior studies have shown that HU values do not differ significantly among lumbar vertebral bodies.^20,26 Additionally, the L3 scanning plane is parallel to the lumbar endplates, L3 experiences relatively balanced loading, which helps reduce bias introduced by biomechanical and anatomical differences between the upper and lower lumbar segments, and it is less affected by common degenerative changes, endplate sclerosis, and focal lesions in the lower lumbar spine. Moreover, prior studies and clinical practice have widely adopted L3 as a stable and representative level for assessing lumbar bone quality; therefore, it can more reliably reflect an individual’s overall bone quality background; therefore, L3 was selected as the target level for analysis.²³ The overall postoperative pedicle screw loosening rate in this study was 14.3% (45/313), consistent with previous studies.²² This study found that age and Goutallier grade were independently associated with the risk of loosening. The number of fused levels was associated with loosening in univariable analysis, and became an independent predictor in a sensitivity analysis after including additional scoliosis cases, details of construct-related effects observed in the expanded cohort (including fusion length and terminal fusion segment) are provided in Supplemental Tables S2 and S4. This finding aligns with prior research: long-segment fusion is one of the most consistently validated non-implant-related risk factors, with the loosening rate increasing as the number of fused levels rises. The primary mechanism is that longer constructs, with greater lever arms and bending moments, concentrate stress on terminal screws, making them more prone to loosening.²⁷

Previous studies suggest that the HU value of a single vertebral body represents the mean HU across multiple local regions within that vertebra, reflecting the vertebra’s overall BMD. However, the HU value of a single vertebral body cannot accurately reflect local bone quality.²⁸ This is further supported by the findings of Xu et al., who noted that relying solely on vertebral body HU is insufficient for a comprehensive assessment of screw loosening risk. They emphasized the need to measure both pedicle HU and vertebral body HU to improve predictive accuracy.¹² Although CT-based HU measurements are convenient and rapid, in patients with low BMD, HU values are typically low, and local differences are difficult to discern. Phenomenologically, the higher the HU in the screw-trajectory region relative to the vertebral body HU, the stronger the bone at the screw site, providing greater support and reducing the risk of screw loosening. Consistent with this, prior studies have shown that patients with lower density along the screw trajectory have significantly higher rates of screw loosening.²⁹ To address this, this study proposed the HU ratio, defined as the ratio of HU in the screw-trajectory region to HU in the vertebral body, to more accurately assess the relative bone quality of the trajectory region across different levels of overall bone quality. Additionally, studies have demonstrated that increased fatty infiltration is characteristic of osteoporotic bone, and the bone marrow fat fraction (BMFF) is closely associated with the onset and progression of osteoporosis, with evidence supporting a clear causal relationship.³⁰ The PBQ score, as an index specifically assessing the degree of marrow fatty infiltration and microstructural status in the pedicle region, further aids in evaluating screw stability.^13,16 Therefore, this study introduced the HU ratio as a novel bone quality predictor, which, in combination with the MRI-based PBQ score, age, and paraspinal muscle fatty infiltration grade, forms a nomogram to assess the risk of pedicle screw loosening after pedicle screw fixation. In line with this, higher Goutallier grades were generally associated with an increased risk of screw loosening, supporting the notion that severe paraspinal muscle fatty degeneration compromises local biomechanical support. The markedly elevated risk estimate observed for Goutallier grade 3 relative to grade 4 in the multivariable model is likely driven by the small sample size in the grade 3 subgroup (n = 10) and the resulting wide confidence interval, and should not be overinterpreted. This model is designed for clinical use, allowing clinicians to input patient-specific variables into the nomogram to obtain a total score and estimate the probability of postoperative pedicle screw loosening.

Our findings demonstrate that higher PBQ scores and lower HU ratios are independently associated with an increased risk of pedicle screw loosening, indicating statistical independence between these two indicators. This independence highlights their complementary roles in assessing bone quality, as they capture different aspects of bone status relevant to screw stability. Additionally, the combined prediction model incorporating these two indicators exhibited excellent discriminative ability and acceptable calibration, further supporting the clinical value of integrating the PBQ score and HU ratio for preoperative risk assessment. These findings align with prior studies, where the loosening group had a significantly higher PBQ score (P < .001).¹⁶ A previous study also reported that L3 vertebral body HU was significantly lower in the loosening group than in the control group (98.6 HU vs 121.4 HU; P < .001),¹² demonstrating that lower HU values are associated with screw loosening, consistent with the present study. Conversely, sex and traditional indices such as DXA were not significant. Furthermore, the PBQ score and HU ratio showed no significant linear or monotonic correlation (Spearman ρ = 0.039, P = .491), suggesting that these two measures capture largely independent aspects of bone quality. Consistent with this, Li et al. observed a significant negative correlation between the PBQ score and quantitative CT BMD (QCT BMD) (r = −0.489; P < .001), indicating that PBQ is not redundant with density-based metrics but instead captures bone status from a complementary perspective.¹⁶ This confirms that the two measures are mechanistically complementary: a higher PBQ score reflects pronounced local marrow fatty infiltration, while a lower HU ratio indicates an imbalance between local and overall mineralization distribution. It is this dual alteration—fatty infiltration coupled with insufficient mineralization—that markedly increases the risk of pedicle screw loosening after pedicle screw fixation.¹⁵ Additionally, our study demonstrated that a combined predictive nomogram constructed using the HU ratio, PBQ score, and other relevant factors identifies high-risk patients more effectively than any single metric. The model offers superior diagnostic performance and clinical applicability. When the predicted probability is ≥0.212, the model shows the greatest net benefit in DCA. Thresholds can and should be flexibly adjusted according to the clinical scenario. Decision-curve analysis demonstrated that our nomogram achieved positive net benefit across a wide range of threshold probabilities. Notably, the threshold of 0.212 coincided with the optimal cutoff determined by the Youden index and our clinical strategy of prioritizing sensitivity to reduce missed high-risk cases, further supporting its clinical practicability. This choice necessarily increases false-positive classification, but the clinical consequence is typically additional surveillance and/or preventive measures rather than irreversible harm, which is acceptable when the cost of a missed loosening event is high. A lower threshold such as 0.212 is suitable for high-risk settings (eg, osteoporosis, long-segment fusion, or high terminal stress) to ensure high sensitivity, whereas a higher threshold can be adopted for low-risk patients with good bone quality and short constructs to minimize unnecessary interventions. Leveraging the mechanistic complementarity of MRI and CT, the combined assessment compensates for the limitations of any single indicator, thereby improving predictive accuracy. Compared to previous studies, our work offers several key advantages. First, the PBQ score specifically targets the pedicle anchorage corridor and reflects the local bone quality microstate at the operative site through the degree of marrow fatty infiltration.¹⁶ Second, the HU ratio corrects for interindividual differences in overall BMD and quantifies the disparity between local and overall BMD, improving comparability. This approach mitigates the overestimation of bone status by DXA in degenerative spines and addresses the limitations of using a single absolute HU value, which does not account for interindividual variability. Given that osteoporotic bone is characterized by marrow adipocyte infiltration and trabecular atrophy, a combined predictive nomogram integrating the HU ratio, PBQ score, and other relevant factors can more precisely assess the preoperative risk of pedicle screw loosening, enabling additional treatment strategies for high-risk patients.¹⁵

This study employed the HU ratio as a key imaging parameter, alongside the PBQ score and clinical variables such as age, and paraspinal muscle fatty infiltration grade, to construct a nomogram for individualized preoperative assessment. All variables are easily obtainable from routine preoperative imaging, are simple to implement with high reproducibility, and incur no additional cost.³¹ The nomogram derived from the model provides a visual estimation of an individual patient’s risk of screw loosening and offers a foundation for personalized preoperative decision-making. For instance, in patients with higher PBQ scores and lower HU ratios, surgeons may consider adjunct fixation devices, cement augmentation, or other supportive measures to reduce the risk of loosening. Additionally, reinforcing nutritional support and rehabilitation management during postoperative follow-up can help mitigate loosening risk and improve outcomes.

In summary, the HU ratio and the nomogram constructed based on the HU ratio and PBQ score not only enhance predictive sensitivity and specificity but also provide scientific evidence for preoperative evaluation and postoperative management in spine surgery, ultimately contributing to improved patient outcomes.

Although this study validated the importance of the HU ratio and the nomogram constructed from the HU ratio and PBQ score for predicting postoperative pedicle screw loosening, it has several limitations. First, it was a single-center retrospective study with only internal validation. Such a design may introduce unpredictable and unavoidable bias due to the non-randomized enrollment of participants, and the definitive conclusions drawn from this study can only be verified by prospective randomized controlled trials (RCTs) in the future. In addition, institutional variability in imaging protocols and surgical techniques may limit the generalizability of our findings; thus, external validation in independent multicenter cohorts is needed to confirm the generalizability of our findings. Second, the HU ratio was primarily measured at L3 (with L2 or L4 used as substitutes when necessary), and this study did not investigate whether optimal thresholds vary across vertebral levels; level-specific effects therefore require further exploration. Finally, the follow-up period was limited to 12 months, which is relatively short to capture late-onset screw loosening. As screw loosening may occur or progress beyond the first postoperative year, longer-term follow-up studies are needed.

Conclusion

This study is the first to introduce the HU ratio to quantify relative mineralization in the anchorage region and confirm its complementarity with the MRI-based PBQ score. A nomogram combining these two measures with clinical variables demonstrated excellent performance, enhancing preoperative risk assessment and guiding interventional decision-making.

Supplemental Material

Supplemental material - Predictive Value of the Preoperative Screw Trajectory-to-Vertebral Body Hounsfield Unit Ratio and the Combined Model Incorporating MRI-Based Pedicle Bone Quality Score for Pedicle Screw Loosening After Pedicle Screw Fixation

Supplemental material for Predictive Value of the Preoperative Screw Trajectory-to-Vertebral Body Hounsfield Unit Ratio and the Combined Model Incorporating MRI-Based Pedicle Bone Quality Score for Pedicle Screw Loosening After Pedicle Screw Fixation by Jinhua Yang, Xincan Wu, Qiancheng Sun, Yunxin Su, Yuanzhen Zhang, Guoyong Yin, and Jian Chen in Global Spine Journal

Footnotes

Acknowledgments

We are grateful to all participants and their families enrolled in this study. We thank for the support of the nursing staffs from the department of orthopaedics of the First Affiliated Hospital of Nanjing Medical University.

ORCID iDs

Jinhua Yang

Xincan Wu

Ethical Considerations

This study was approved by the Ethics Committee of the First Affiliated Hospital of Nanjing Medical University (Approval No. 2025-SR-769). The study was conducted in accordance with the Declaration of Helsinki.

Consent to Participate

This retrospective study was approved by the Ethics Committee of the First Affiliated Hospital of Nanjing Medical University (Approval No. 2025-SR-769). Informed consent was waived by the Ethics Committee due to the retrospective nature of the study and the use of anonymized clinical data.

Consent for Publication

Not applicable. The images used in this study (X-ray, CT scans, etc.) do not contain any directly identifiable information. Written informed consent for publication was not required as per the approval from the Ethics Committee of the First Affiliated Hospital of Nanjing Medical University (Approval No. 2025-SR-769).

Author Contributions

Jinhua Yang and Xincan Wu contributed equally to this work. Jinhua Yang and Xincan Wu were responsible for study conception and design, data collection, and manuscript drafting. Qiancheng Sun and Yunxin Su contributed to data acquisition and data analysis. Yuanzhen Zhang provided statistical analysis and technical support. Guoyong Yin and Jian Chen supervised the study, critically revised the manuscript for important intellectual content, and served as corresponding authors. All authors reviewed and approved the final manuscript.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the High-level Talent Cultivation Program (Phase 1) - A Category Fund (Project No. CZ0121002010037, EV24) and the High-level Talent Cultivation Program (Phase 1) - C Category Fund (Project No. CZ0121002010037, EV24) of the First Affiliated Hospital of Nanjing Medical University.

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data Availability Statement

The data supporting the findings of this study are available upon reasonable request from the corresponding author. Due to privacy concerns, the data cannot be shared publicly.*

Supplemental Material

Supplemental material for this article is available online.

References

Yang

Kong

, et al. Unilateral versus bilateral pedicle screw fixation for treating two-level lumbar degenerative diseases. Sci Rep. 2025;15(1):27340. doi:10.1038/s41598-025-13015-1

Marie-Hardy

Pascal-Moussellard

Barnaba

Bonaccorsi

Scemama

. Screw loosening in posterior spine fusion: prevalence and risk factors. Glob Spine J. 2020;10(5):598-602. doi:10.1177/2192568219864341

Löffler

Sollmann

Burian

, et al. Opportunistic osteoporosis screening reveals low bone density in patients with screw loosening after lumbar semi-rigid instrumentation: a case-control study. Front Endocrinol. 2021;11:552719. doi:10.3389/fendo.2020.552719

Zhu

Hua

, et al. Vertebral bone quality score as a predictor of pedicle screw loosening following surgery for degenerative lumbar disease. Spine. 2023;48(23):1635-1641. doi:10.1097/BRS.0000000000004577

Bloem

Reijnierse

Huizinga

TWJ

Van Der Helm-van Mil

AHM

. MR signal intensity: staying on the bright side in MR image interpretation. RMD Open. 2018;4(1):e000728. doi:10.1136/rmdopen-2018-000728

on behalf of the Scientific Advisory Board of the European Society for Clinical and Economic Aspects of Osteoporosis (ESCEO) and the Committees of Scientific Advisors and National Societies of the International Osteoporosis Foundation (IOF) Kanis

Cooper

Rizzoli

Reginster

. European guidance for the diagnosis and management of osteoporosis in postmenopausal women. Osteoporos Int. 2019;30(1):3-44. doi:10.1007/s00198-018-4704-5

Yang

Wang

Song

. Novel MRI‐based pedicle bone quality score independently predicts pedicle screw loosening after degenerative lumbar fusion surgery. Orthop Surg. 2024;16:2372-2379. doi:10.1111/os.14146

Grams

Rehwald

Bartsch

, et al. Correlation between degenerative spine disease and bone marrow density: a retrospective investigation. BMC Med Imag. 2016;16(1):17. doi:10.1186/s12880-016-0123-2

Bao

Zou

. Characteristics of the DXA measurements in patients undergoing lumbar fusion for lumbar degenerative diseases: a retrospective analysis of over 1000 patients. CIA. 2021;16:1131-1137. doi:10.2147/CIA.S300873

10.

Schreiber

Anderson

Rosas

Buchholz

. Hounsfield units for assessing bone mineral density and strength: a tool for osteoporosis management. J Bone Joint Surg. 2011;93(11):1057-1063. doi:10.2106/JBJS.J.00160

11.

Andresen

Schröder

Haider

Schober

Andresen

. Osteoporosis assessment using bone density measurement in hounsfield units in the femoral native CT cross-section: a comparison with computed tomography X-ray absorptiometry of the hip. Diagnostics. 2025;15(8):1014. doi:10.3390/diagnostics15081014

12.

Zou

, et al. Hounsfield units of the vertebral body and pedicle as predictors of pedicle screw loosening after degenerative lumbar spine surgery. Neurosurg Focus. 2020;49(2):E10. doi:10.3171/2020.5.FOCUS20249

13.

Zhang

Xie

Song

Zeng

. The preoperative hounsfield unit value at the position of the future screw insertion is a better predictor of screw loosening than other methods. Eur Radiol. 2022;33(3):1526-1536. doi:10.1007/s00330-022-09157-9

14.

Gleich

Steiner

Ehrnthaller

, et al. CT-based evaluation of hounsfield units—A novel screening tool for undiagnosed osteoporosis in patients with fragility fractures of the pelvis. JCM. 2025;14(10):3346. doi:10.3390/jcm14103346

15.

Jiang

, et al. Prediction of screw loosening after dynamic pedicle screw fixation with lumbar polyetheretherketone rods using magnetic resonance imaging-based vertebral bone quality score. Neurospine. 2024;21(2):712-720. doi:10.14245/ns.2448184.092

16.

Yang

Wang

Song

. MRI-based pedicle bone quality score: correlation to quantitative computed tomography bone mineral density and its role in quantitative assessment of osteoporosis. Spine J. 2024;24(10):1825-1832. doi:10.1016/j.spinee.2024.06.007

17.

Spirig

Sutter

Götschi

Farshad-Amacker

Farshad

. Value of standard radiographs, computed tomography, and magnetic resonance imaging of the lumbar spine in detection of intraoperatively confirmed pedicle screw loosening—A prospective clinical trial. Spine J. 2019;19(3):461-468. doi:10.1016/j.spinee.2018.06.345

18.

Zou

Muheremu

Sun

Zhong

Jiang

. Computed tomography hounsfield unit–based prediction of pedicle screw loosening after surgery for degenerative lumbar spine disease. J Neurosurg Spine. 2020;32(5):716-721. doi:10.3171/2019.11.SPINE19868

19.

Sun

Jiang

, et al. A study of screw placement to obtain the optimal pull-out resistance of lumbar pedicle screws—Analysis of hounsfield units measurements based on computed tomography. BMC Musculoskelet Disord 2022;23(1):124. doi:10.1186/s12891-022-05074-6

20.

Feng

Zhou

Qiu

Jiang

. The influence of proliferative tissue on hounsfield unit and its correlation with BMD in middle-aged and elderly patients with lumbar degenerative diseases. J Orthop Surg Res. 2024;19(1):623. doi:10.1186/s13018-024-05130-y

21.

Zhao

Wang

, et al. Three-dimensional hounsfield units measurement of pedicle screw trajectory for predicating screw loosening in lumbar fusion surgery. CIA. 2023;18:485-493. doi:10.2147/CIA.S389059

22.

Ehresman

Pennington

Schilling

, et al. Novel MRI-based score for assessment of bone density in operative spine patients. Spine J. 2020;20(4):556-562. doi:10.1016/j.spinee.2019.10.018

23.

Pickhardt

Pooler

Lauder

Del Rio

Bruce

Binkley

. Opportunistic screening for osteoporosis using abdominal computed tomography scans obtained for other indications. Ann Intern Med. 2013;158(8):588-595. doi:10.7326/0003-4819-158-8-201304160-00003

24.

Lewiecki

Gordon

Baim

, et al. International society for clinical densitometry 2007 adult and pediatric official positions. Bone. 2008;43(6):1115-1121. doi:10.1016/j.bone.2008.08.106

25.

Zhan

Wei

Guo

, et al. The predictive value of endplate morphology and pedicle screw bone quality score on screw loosening after single-level lumbar spinal fusion surgery. J Orthop Surg Res. 2024;19(1):898. doi:10.1186/s13018-024-05367-7

26.

Yamato

Sato

Togawa

, et al. Differences in the geometrical spinal shape in the sagittal plane according to age and magnitude of pelvic incidence in healthy elderly individuals. J Orthop Sci. 2020;25(4):557-564. doi:10.1016/j.jos.2019.07.005

27.

Yuan

Zhang

Zeng

Chen

. Incidence, risk, and outcome of pedicle screw loosening in degenerative lumbar scoliosis patients undergoing long-segment fusion. Glob Spine J. 2023;13(4):1064-1071. doi:10.1177/21925682211017477

28.

Chen

Zheng

Wang

. Hounsfield unit for assessing bone mineral density distribution within lumbar vertebrae and its clinical values. Front Endocrinol. 2024;15:1398367. doi:10.3389/fendo.2024.1398367

29.

Zhang

Huang

Dong

. Correlation analysis of BMD in different regions of the vertebrae determined by QCT and DXA on pedicle screw loosening. Sci Rep. 2025;15(1):11850. doi:10.1038/s41598-025-91816-0

30.

Mesa-Eguiagaray

Morris

, et al. Clinical implications of bone marrow adiposity identified by phenome-wide association and mendelian randomization in the UK biobank. Nat Commun. 2025;16(1):8332. doi:10.1038/s41467-025-63395-1

31.

Liang

Chen

, et al. Comparison of the predictive roles of CT- and MRI-based pedicle regional osteoporosis status measurements for pedicle screw loosening after posterior lumbar interbody fusion. Glob Spine J. 2025;16(2):1075-1084. doi:10.1177/21925682251371602. Published online.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.47 MB