Opportunistic screening of osteoporosis in lung transplant recipients: diagnostic value of pre-transplant thoracic CT using vertebral Hounsfield units

Abstract

Purpose

To evaluate the diagnostic performance of pre-transplant chest computed tomography (CT) as an opportunistic screening tool for predicting osteoporosis 1 year after lung transplantation.

Method

We retrospectively enrolled lung transplant patients from a tertiary center in Zhejiang, China, between September 2022 and December 2023. Standardized regions of interest were placed on the first thoracic vertebra (T1) through the first lumbar vertebra (L1) to measure CT-derived Hounsfield unit (HU) values. Patients were categorized into normal BMD, osteopenia, and osteoporosis groups on the basis of their baseline dual-energy X-ray absorptiometry (DEXA) T-scores. Clinical baseline characteristics and HU values were compared among the three groups. The correlation between pre-transplant HU values and DEXA T-scores was analyzed, and receiver operating characteristic (ROC) curves were used to assess the diagnostic performance of pre-transplant HU values in predicting post-transplant osteopenia and osteoporosis.

Results

A total of 139 lung transplant patients were included in the study, among whom 60 had normal bone mass, 51 had osteopenia, and 28 had osteoporosis at baseline. Notably, HU values significantly differed among the three groups (P < 0.05). The HU values of T1–L1 were weakly to moderately positively correlated with the T-scores of the lumbar spine (LS), femoral neck (FN), and total hip (TH; r = 0.28–0.51, P < 0.01). After lung transplantation, the proportion of patients with low bone mineral density (BMD) increased, and T-scores of the FN and TH decreased (P < 0.01). Baseline HU values showed good diagnostic performance for detecting bone abnormalities 1 year after transplantation, with T2 showing particular potential for distinguishing osteoporosis from non-osteoporosis and T9 for distinguishing low BMD from normal BMD.

Conclusions

Pretransplant thoracic CT provides a clinically feasible and opportunistic screening approach to stratify osteoporosis risk in lung transplant candidates without additional radiation exposure. By quantifying T2 and T9 vertebral HU values, clinicians can identify high-risk patients for timely intervention, potentially reducing fracture-related morbidity.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13018-025-06271-4.

Introduction

Lung transplantation is an effective therapeutic option for patients with end-stage lung disease, offering significantly improved survival outcomes [1]. With prolonged survival, certain potential complications have increasingly drawn attention, notably low bone mineral density (BMD), including osteopenia and osteoporosis. The prevalence of osteoporosis among lung transplant candidates is reportedly as high as 32–61%, which further increases to 40–78% after transplantation, primarily due to immunosuppressive therapy and prolonged physical inactivity [2–5]. Low BMD significantly elevates the risk of fragility fractures, adversely affecting patients’ quality of life, increasing healthcare costs, and contributing to higher mortality [6–8]. Thus, early screening of BMD to identify high-risk patients and implementing timely preventive interventions are essential strategies for reducing postoperative adverse outcomes [9].

Currently, dual-energy X-ray absorptiometry (DEXA) is widely regarded as the gold standard for osteoporosis screening [10–12]. The International Society for Heart and Lung Transplantation (ISHLT) recommends routine DEXA assessments for lung transplant patients, both preoperatively and at 1 year after transplantation [11, 13, 14]. However, DEXA, as a two-dimensional imaging modality, has inherent limitations. Its measurements can be significantly influenced by confounding factors, such as spinal degeneration, vertebral deformities, aortic calcification, and abdominal adiposity, reducing diagnostic sensitivity [13]. In addition, incomplete DEXA data may arise from preoperative physical limitations or insufficient adherence to postoperative follow-up, posing challenges for accurate bone health assessment and potentially delaying early intervention in patients at high risk for osteoporosis.

In recent years, BMD assessment based on computed tomography (CT)-derived Hounsfield units (HU) has gained increasing attention. Unlike DEXA, CT, as a three-dimensional imaging modality, allows precise quantification of volumetric bone mineral density within vertebral trabecular regions by effectively excluding interference from marginal sclerosis, calcifications, and other confounding factors, thus offering superior accuracy [14–16]. Importantly, lung transplant patients routinely undergo chest CT examinations as part of preoperative evaluations and postoperative follow-up, which allows for directly leveraging existing imaging data to assess BMD, thus minimizing additional radiation exposure and examination costs while addressing DEXA data gaps.

Previous studies have shown a significant positive correlation between vertebral HU values and DEXA-derived BMD, with HU values also showing good accuracy in diagnosing osteoporosis [17]. For instance, Luo et al. [18] reported that mid-spinal CT scans effectively identified patients at high risk for osteoporosis. Furthermore, Pinto et al. [19] confirmed that HU values have strong diagnostic performance in detecting osteoporosis, with the area under the receiver operating characteristic (ROC) curve ranging from 0.66 to 0.96, highlighting the potential clinical utility of CT-based bone density assessment. Despite the established correlations between HU values and DEXA scan findings in general populations, lung transplant recipients are a unique cohort given their chronic glucocorticoid use, prolonged immobilization, and immunosuppression [20]. The applicability of CT-based osteoporosis screening in this high-risk group remains underexplored.

Therefore, in this study, we aim to investigate the correlation between pretransplant CT-derived vertebral HU values and DEXA-measured BMD in lung transplant patients and to evaluate the diagnostic utility of pretransplant HU values for osteoporosis prediction at 1 year after transplantation. Through this research, we seek to establish a reliable and convenient supplemental approach for assessing bone density, facilitating timely identification and targeted intervention in high-risk patients and ultimately improving bone health outcomes and quality of life in lung transplant recipients.

Methods

Ethical approval

The study was conducted in accordance with the Declaration of Helsinki (as revised in 2013). The study was approved by the Ethics Committee at the Second Affiliated Hospital, Zhejiang University School of Medicine (IRB no. 2024 − 1404).

Study design and population

This was a retrospective study conducted in the Lung Transplantation Department of the Second Affiliated Hospital, Zhejiang University School of Medicine. The study included all patients who underwent lung transplantation at our center between September 2022 and December 2023. We excluded patients (1) who underwent re-transplantation, (2) who were aged < 18 years, (3) who had incomplete clinical data, (4) who had vertebral structural abnormalities like compression fractures or severe osteophytes, (5) whose chest CT scans did not completely cover the first thoracic vertebra (T1) through the first lumbar vertebra (L1), and; (6) who had incomplete imaging data, particularly lacking pretransplant chest CT scans, baseline pretransplant DEXA measurements, or 12-month post-transplant DEXA measurements. All clinical data were obtained from the hospital’s electronic medical record system.

Dual-energy X-ray absorptiometry

Areal BMD (g/cm²) was measured using DEXA (Hologic Discovery A, Hologic Inc., Bedford, MA, USA) at three anatomical sites: the lumbar spine (LS, mean of L1–L4), femoral neck (FN), and total left hip (TH). All DEXA scans were performed by certified technologists following manufacturer protocols, with daily calibration using phantom standards. To minimize operator-dependent variability, the same DEXA machine and software (APEX v5.6) were used for all participants. DEXA results were interpreted according to the 1994 WHO diagnostic criteria: (1) normal: T-score ≥ − 1.0, (2) osteopenia: −2.5 < T-score < − 1.0; (3) osteoporosis: T-score ≤ − 2.5 [21].

CT acquisition and analysis

CT scans were performed using a third-generation dual-source CT system (Siemens SOMATOM New Force, Siemens Healthineers). All patients were imaged during full inspiration, with the scanning direction extending from the diaphragm’s base to the pulmonary apex. The technical parameters were as follows: acquisition collimation, 192 mm × 0.6 mm; rotation time, 0.25 s; pitch, 2; slice thickness, 1.5 mm; increment, 1.5 mm; kernel, Br40; and ADMIRE 3. The tube voltage was set at 120 kV, with a reference tube current of 80 mAs. The automatic tube current modulation technique (CareDose 4D, Siemens Healthcare) was consistently used across all examinations. Mediastinal window images (window width, 350; window level, 40) were transferred to a dedicated workstation (Syngo.via, version VB40, Siemens Healthineers) for postprocessing.

Vertebral levels from T1 to L1 were systematically identified on coronal reformatted images to ensure adequate image quality for subsequent quantitative analysis. Each measurement was always obtained from the same anatomically identified vertebral level for all patients to ensure consistency and comparability. Region of interest (ROI) placement was performed manually on axial images. ROIs were circular or elliptical, with areas adjusted between 50 mm² and 150 mm² according to individual vertebral body size. Meticulous care was taken to avoid the cortical bone, the basivertebral vein, areas of calcification, and any vertebral bodies showing fractures or deformities. The placement adhered strictly to a standardized protocol, positioning the ROI centrally within the trabecular bone compartment of each vertebral body and maximizing its area where feasible. Multiplanar reconstruction (MPR) images in sagittal, coronal, and axial planes were used for three-dimensional verification of ROI placement. This ensured that the ROIs resided entirely within the trabecular bone, devoid of cortical bone inclusion or other structural abnormalities. Vertebral bodies showing evidence of bone destruction, hemangioma, focal bone marrow hyperplasia, or other significant pathologies were excluded from analysis (Fig. 1). Subsequently, the mean bone density within each ROI was automatically calculated using dedicated CT image analysis software and recorded in Hounsfield units (HU). For each vertebral body, the final HU value used in statistical analysis was derived as the mean of the measurements obtained by two independent physicians. Inter-observer agreement for HU measurements across all individual vertebral levels (T1–L1) was assessed using the intraclass correlation coefficient (ICC). The analysis showed excellent agreement, with ICC values exceeding 0.80 for all 13 vertebral levels.

Fig. 1.

Pre-transplant CT evaluation of a 47-year-old woman with lymphangioleiomyomatosis (LAM). (A) Axial lung window CT image; (B) three-dimensional pulmonary reconstruction; (C) measurement of bone CT attenuation value at the L2 vertebral level; (D) sagittal view illustrating vertebral labeling

Statistical analyses

Categorical variables were presented as frequencies and percentages. Normally distributed continuous variables were expressed as mean ± standard deviation, and non-normally distributed ones were described as median and interquartile range. Differences in continuous variables among multiple groups were analyzed using ANOVA for normally distributed data or using the Kruskal–Wallis test for non-normally distributed data. The least significant difference (LSD) method was applied for pairwise comparisons between groups. Categorical variables were compared using the chi-square test, with pairwise comparisons further conducted by the partitioning the chi-square method and Bonferroni correction, resulting in an adjusted significance level of α = 0.05/3 = 0.016. Spearman correlation analysis was used to assess relationships of HU values with T-scores, age, and BMI. Paired-samples t-test was performed to compare T-scores before transplantation and 1 year after transplantation. ROC curve analysis was used to evaluate the diagnostic value of CT-based HU values for osteoporosis or low BMD (normal and osteopenia) 1 year after transplantation. Additionally, a supplementary multi-class ROC analysis was performed to assess three-category classification performance(normal/osteopenia/osteoporosis), with results evaluated using both micro-average and macro-average AUCs. A two-sided P value of < 0.05 was considered statistically significant. Statistical analyses were performed using SPSS version 27.0, figures were generated with Origin 2024, and multi-class ROC analysis was conducted using R v4.1.1 (R Foundation for Statistical Computing).

Results

The differences in clinical baseline data and HU values among normal BMD, osteopenia, and osteoporosis groups

A total of 226 lung transplant recipients were initially screened for eligibility. After applying the exclusion criteria, 87 patients were excluded. Thus, 139 patients were finally included in the analysis. The detailed flowchart of patient selection is presented in Fig. 2. Among these patients, 60 patients (43.17%) had normal BMD at baseline, 51 patients (36.70%) had osteopenia, and 28 patients (20.14%) had osteoporosis. Analysis of baseline clinical characteristics revealed significant differences among the three groups in terms of sex, age, BMI, indication for transplantation, history of diabetes, and glucocorticoid use (P < 0.05; Table 1).

Fig. 2.

Study participant selection flowchart

Table 1.

Baseline demographic and clinical characteristics of lung transplant patients

Variables	Normal	Osteopenia	Osteoporosis	P1	P2	P3	P4
Male (n, %)	49 (81.7%)	49 (96.1%)	20 (71.4%)	0.009	0.019	0.277	0.003
Age (years)	57 (43.25, 64.75)	62 (57, 67)	57.5 (51.25, 63)	0.002	0.002	0.999	0.040
BMI (kg/m2)	21.319 ± 4.153	21.298 ± 3.719	18.671 ± 3.476	0.007	0.978	0.003	0.005
Transplant indication (n, %)				0.002	0.003	0.001	0.332
Pneumoconiosis	5 (8.3%)	10 (19.6%)	6 (21.4%)
Interstitial lung disease	36 (60%)	22 (43.1%)	7 (25%)
COPD	8 (13.3%)	17 (33.3%)	12 (42.9%)
Others	11 (18.3%)	2 (3.9%)	3 (10.7)
History of diabetes mellitus (n, %)	9 (15.0%)	12 (23.5%)	3 (10.7%)	0.293	N/A	N/A	N/A
History of hypertension (n, %)	10 (16.7%)	12 (23.5%)	2 (7.1%)	0.180	N/A	N/A	N/A
Ex-smoker (n, %)	34 (56.7%)	36 (70.6%)	15 (53.6%)	0.213	N/A	N/A	N/A
Use of GCs (n, %)	15 (25.0%)	23 (45.1%)	13 (46.4%)	0.044	0.013	0.232	0.117
Bone-active drug use (n, %)	13 (21.7%)	14 (27.5%)	11 (39.3%)	0.225	N/A	N/A	N/A
Calcium (mmol/L)	2.29 (2.20, 2.39)	2.30 (2.23, 2.42)	2.27 (2.20, 2.39)	0.435	N/A	N/A	N/A
TSH (mIU/L)	1.33 (0.79, 1.98)	1.55 (1.18, 2.16)	1.23 (0.82, 2.01)	0.356	N/A	N/A	N/A
The CT values of thoracic and the first lumbar vertebrae
T1	176.68 ± 43.91	145.73 ± 34.15	139.92 ± 43.61	< 0.001	< 0.001	< 0.001	0.543
T2	173.11 ± 40.04	135.98 ± 33.58	134.53 ± 42.99	< 0.001	< 0.001	< 0.001	0.570
T3	166.98 ± 42.72	135.92 ± 33.58	128.21 ± 43.95	< 0.001	< 0.001	< 0.001	0.413
T4	157.81 ± 43.94	126.11 ± 35.26	123.39 ± 44.00	< 0.001	< 0.001	< 0.001	0.778
T5	150.10 ± 41.87	118.13 ± 31.81	120.05 ± 45.94	< 0.001	< 0.001	0.001	0.836
T6	143.15 ± 39.64	110.55 ± 33.77	110.85 ± 39.47	< 0.001	< 0.001	< 0.001	0.973
T7	134.77 ± 40.96	103.51 ± 32.33	106.20 ± 37.90	< 0.001	< 0.001	0.001	0.760
T8	134.38 ± 39.92	100.94 ± 33.73	101.75 ± 39.60	< 0.001	< 0.001	< 0.001	0.927
T9	136.09 ± 38.68	99.79 ± 31.87	97.64 ± 43.23	< 0.001	< 0.001	< 0.001	0.806
T10	140.43 ± 37.75	103.30 ± 34.55	102.70 ± 42.33	< 0.001	< 0.001	< 0.001	0.994
T11	130.05 (105.83, 173.12)	102.50 (79.60, 116.40)	91.45 (68.90, 144.92)	< 0.001	< 0.001	0.001	0.999
T12	132.54 ± 40.65	96.27 ± 34.67	97.56 ± 37.96	< 0.001	< 0.001	< 0.001	0.877
L1	120.22 (101.15, 161.24)	89.40 (76.90, 107.50)	85.85 (59.30, 127.86)	< 0.001	< 0.001	< 0.001	0.999

Abbreviations: BMI, body mass index; COPD, chronic obstructive pulmonary disease; T1–L1, all thoracic vertebrae and the first lumbar vertebra; GCs, glucocorticoids; TSH, thyroid-stimulating hormone; P1, overall differences among the three groups; P2, normal group vs. osteopenia group; P3, normal group vs. osteoporosis group; P4, osteopenia group vs. osteoporosis group; N/A, not applicable

Pairwise comparisons of categorical variables were performed using the chi-square partition method. However, for the use of glucocorticoid, a combined grouping approach was used: normal vs. abnormal group (osteopenia + osteoporosis): P2 = 0.013; normal + osteopenia vs. osteoporosis: P3 = 0.232; normal + osteoporosis vs. osteopenia: P4 = 0.117

The HU values of T1–L1 vertebrae significantly differed among the three groups (P < 0.05). In particular, the normal BMD group had significantly higher HU values than osteopenia and osteoporosis groups (P < 0.05). In addition, starting from T1, the HU values in all three groups showed a declining trend, reaching an initial trough at T7–T9, followed by a marginal increase at T10–T11, and finally dropping to the lowest value at L1 (Table 1; Fig. 3).

Fig. 3.

Differences in the distribution of HU values of T1–L1 among normal BMD, osteopenia, and osteoporosis groups. ***:P < 0.001

The correlation of age, body mass index, and T-score with HU values

Spearman correlation analysis revealed that HU values of T1–L1 were weakly to moderately positively correlated with T-scores of the LS, FN, and TH (r = 0.28–0.51, P < 0.05), with the strongest correlation observed between HU values of T2 and the T-score of LS (r = 0.51, P < 0.05), followed by that between HU values of T9 and the T-score of LS (r = 0.50, P < 0.05). In addition, the correlation of the HU value of T2 was higher with T-scores of FN and TH than of other vertebrae. HU values were significantly negatively correlated with age, indicating weak to moderate correlations (r = -0.31 to -0.54, P < 0.05), most notably at T10, T12, and L1 (r = -0.54 to -0.50). In addition, the HU values of T1, T2, T11, and L1 showed a weak negative correlation with body mass index (BMI; r = -0.16 to -0.27, P < 0.05; Supplementary Fig. 1).

Changes in bone status and BMD before and after transplantation

One year after lung transplantation, 51 patients (36.69%) had normal BMD, 56 patients (40.29%) had osteopenia, and 32 patients (23.02%) had osteoporosis. As shown in Fig. 4, most patients maintained a stable bone status during the first postoperative year, with few patients experiencing changes. In particular, 2 patients (3.33%) with normal BMD and 7 patients (13.73%) with osteopenia at baseline progressed to osteoporosis at 1 year after transplantation. However, as shown in Table 2, T-scores of the FN and TH were significantly lower at 1 year after transplantation compared to baseline (P < 0.001), indicating a continued decline in BMD after transplantation.

Fig. 4.

Transition of the BMD status from baseline to 1 year after lung transplantation

Table 2.

Longitudinal changes in T-scores before and after lung transplantation

	Baseline	Post-transplant	P
LS	-0.601 ± 1.507	-0.617 ± 1.372	0.768
FN	-1.131 ± 1.083	-1.342 ± 1.087	< 0.001
TH	-1.001 ± 1.089	-1.168 ± 1.148	< 0.001

Abbreviations: LS, lumbar spine (L1–L4); FN, femoral neck; TH, total left hip

Diagnostic value of CT-derived HU values for osteopenia and osteoporosis at 1 year after lung transplantation

Our findings revealed that HU values of each vertebral level from T1 to L1 showed considerable diagnostic performance for assessing bone status 1 year after lung transplantation. For differentiating low BMD (including osteopenia and osteoporosis) from normal BMD, the AUC of each vertebral segment ranged from 0.709 to 0.751, with sensitivities of 55.7–80.7% and specificities of 60.8–82.4%. Notably, the T9 vertebra achieved the highest AUC (0.751) with an optimal cutoff of 113.05 HU, yielding a sensitivity of 64.8%, a specificity of 74.5%, and an overall accuracy of 68.35% for distinguishing low BMD from normal BMD. For distinguishing osteoporosis from non-osteoporosis (including individuals with normal BMD or osteopenia), the AUC ranged from 0.650 to 0.698, with sensitivities of 46.9–68.8% and specificities of 61.7–83.2%. Among these, the T2 vertebra exhibited the highest AUC (0.698) with an optimal cutoff of 137.20 HU, resulting in a sensitivity of 68.8%, a specificity of 69.2%, and an overall accuracy of 69.06% in differentiating osteoporosis from non-osteoporosis. Furthermore, we also assessed the diagnostic performance of HU values for distinguishing between osteopenia and osteoporosis, providing additional reference for clinical evaluation. Relevant data are presented in Supplementary Table 1. To provide a more comprehensive evaluation of the diagnostic performance of HU values, we conducted a multi-class ROC analysis based on all measured vertebral levels. The analysis yielded a macro-average AUC of 0.65 and a micro-average AUC of 0.64, indicating moderate discriminative ability for three-category classification (Supplementary Table 2 and Supplementary Fig. 2). These findings support the utility of preoperative CT-based HU measurements in stratifying bone density risk among lung transplant recipients (Table 3; Fig. 5).

Table 3.

ROC analysis of HU values for distinguishing low BMD and osteoporosis at 1 year after lung transplantation

	Distinguishing low BMD from normal				Distinguishing osteoporosis from non-osteoporosis
	AUC (95% CI)	Optimal threshold (HU)	Sensitivity, %	Specificity, %	AUC (95% CI)	Optimal threshold (HU)	Sensitivity, %	Specificity, %
T1	0.719 (0.629, 0.809)	157.76	63.6%	74.5%	0.664 (0.556, 0.771)	127.30	46.9%	83.2%
T2	0.736 (0.654, 0.819)	137.70	55.7%	82.4%	0.698 (0.590, 0.806)	137.20	68.8%	69.2%
T3	0.726 (0.641, 0.812)	155.25	70.5%	68.6%	0.686 (0.578, 0.791)	132.1	65.6%	72.9%
T4	0.740 (0.655, 0.825)	143.10	70.5%	72.5%	0.686 (0.581, 0.791)	119.65	65.6%	69.2%
T5	0.709 (0.619, 0.825)	140.00	73.9%	64.7%	0.665 (0.551, 0.779)	120.45	68.8%	67.3%
T6	0.714 (0.625, 0.804)	124.90	69.3%	68.6%	0.682 (0.576, 0.788)	103.55	62.5%	74.8%
T7	0.734 (0.647, 0.820)	121.6	71.6%	66.7%	0.664 (0.557, 0.770)	97.35	62.5%	73.8%
T8	0.727 (0.639, 0.816)	128.30	80.7%	60.8%	0.648 (0.544, 0.752)	96.9	59.4%	68.2%
T9	0.751 (0.668, 0.834)	113.05	64.8%	74.5%	0.678 (0.568, 0.787)	97.05	62.5%	72.9%
T10	0.749 (0.664, 0.834)	110.40	58.0%	80.4%	0.650 (0.543, 0.757)	96.33	53.1%	76.6%
T11	0.743 (0.656, 0.829)	105.38	56.8%	82.4%	0.653 (0.546, 0.760)	104.3	65.6%	66.4%
T12	0.744 (0.657, 0.830)	117.80	72.7%	64.7%	0.657 (0.551, 0.764)	85.75	50.0%	80.4%
L1	0.738 (0.650, 0.826)	100.45	61.4%	78.4%	0.657 (0.548, 0.767)	98.25	65.6%	61.7%

Low BMD represents both osteopenia and osteoporosis; non-osteoporosis represents normal BMD and osteopenia

Abbreviations: T1–L1, all thoracic vertebra and the first lumbar vertebra; 95% CI, 95% confidence interval; AUC, area under the ROC curve

Abbreviations: T1–L1, all thoracic vertebrae and the first lumbar vertebra

Fig. 5.

ROC analysis of HU values for distinguishing low BMD and osteoporosis at 1 year after lung transplantation. (A) Distinguishing low BMD (osteopenia + osteoporosis) from normal BMD. (B) Distinguishing osteoporosis from non-osteoporosis (normal + osteopenia)

Discussion

This study is the first to investigate the correlation between baseline HU values and DEXA-derived T-scores of the LS, FN, and TH in lung transplant recipients. We also investigated the diagnostic performance of HU values for bone status 1 year after transplantation. The main findings are as follows. Significant differences in CT values of T1–L1 were observed among the three groups with normal BMD, osteopenia, and osteoporosis at baseline (P < 0.01). HU values of T1–L1 showed weak to moderate positive correlation with the T-scores of LS, FN, and TH (R = 0.28–0.51, P < 0.01). After lung transplantation, the proportion of patients with low BMD increased, and T-scores of the FN and TH decreased (P < 0.01). Baseline HU values showed good diagnostic performance for detecting bone abnormalities 1 year after transplantation, with T2 showing particular potential for distinguishing osteoporosis from non-osteoporosis and T9 for distinguishing low BMD from normal BMD.

Pickhardt et al. [22] and Yang et al. [17] have demonstrated in the general population that vertebral HU values decrease progressively from normal BMD to osteopenia and then to osteoporosis. Similarly, in the lung transplant population, we also observed a significant stepwise decrease in HU values of T1–L1 vertebrae among the three groups (P < 0.05), which is highly consistent with the previously reported pattern. Spearman correlation analysis revealed that HU values of T1–L1 showed weak to moderate positive correlation with the T-scores of the LS, FN, and TH (r = 0.28–0.51, P < 0.01), further confirming the consistency between CT-derived HU values and DEXA-measured BMD. Although this relationship has been established in previous studies, to our knowledge, our study is the first to focus on lung transplant recipients, indicating that preoperative chest CT values can help assess BMD in this specific population and provide empirical evidence for opportunistic screening and early identification of osteopenia and osteoporosis.

Our findings showed an overall declining trend in BMD after lung transplantation. Currently, the European Calcified Tissue Society (ECTS) advocates for early identification of high-risk patients during the lung transplant waiting period and prompt initiation of appropriate interventions [23, 24]. In this study, we further evaluated the predictive value of preoperative chest CT HU values of T1–L1 vertebrae for bone status at 1 year after lung transplantation, with the aim to provide a reference for the early identification of high-risk patients. In particular, the T9 vertebra showed relatively favorable diagnostic performance for distinguishing low bone mass (including osteopenia and osteoporosis) from normal bone mass (AUC = 0.751, cutoff value 113.05 HU, specificity 74.5%, sensitivity 64.8%). For differentiating osteoporosis from non-osteoporosis (including normal bone mass and osteopenia), the T2 vertebra showed relatively higher accuracy (AUC = 0.698, cutoff value 137.20 HU, specificity 69.2%, sensitivity 68.8%). These results suggest that the T2 and T9 vertebrae have value for opportunistic screening in lung transplant recipients and are useful references for preoperative assessment of bone status.

Compared with previous studies in the general population, the optimal vertebral segments and corresponding HU thresholds recommended by our study show notable differences. For example, Yang et al. [17] identified T5 as the most discriminative vertebra for distinguishing between osteopenia and osteoporosis, with cutoff values of 176.5 HU and 150.5 HU, respectively, whereas Pan et al. [25] proposed a threshold of HU ≤ 157 for identifying low BMD. The reasons for these differences are multifactorial. Lung transplant candidates often require long-term immunosuppressive therapy (e.g., glucocorticoids and calcineurin inhibitors) before transplantation, leading to extensive trabecular bone loss and low HU values when compared with the general population [26, 27]. In addition, the study has shown that vertebral fractures after lung transplantation are more frequently observed at T6, T7, T11, and T12, indicating that the high-risk regions for fractures in lung transplant recipients are more widely distributed and not limited to the conventional thoracolumbar junction [28, 29]. This distribution pattern may be related to chronic respiratory dysfunction in lung transplant patients, which leads to altered thoracic biomechanics and subsequently affects the loading patterns of different spinal segments, resulting in a shift in the distribution of bone loss. Conversely, differences in CT scanner detector types, scanning protocols, and image reconstruction algorithms may also contribute to the variability in HU values between studies [30].

The correlation analysis in this study showed that HU values of T2 and T9 vertebrae had the strongest association with baseline LS T-scores, suggesting that CT values at these segments can better reflect DEXA-measured bone mineral density. Although routine chest CT scans cover vertebrae from T1 to T12, in clinical practice, the image quality is generally best in the upper and mid-thoracic segments, making T2 and T9 particularly advantageous for practical application [31]. Therefore, our findings suggest that HU values of T2 and T9 can serve as a useful supplement for preoperative screening of bone abnormalities in lung transplant recipients. In practice, radiologists or clinicians can routinely assess the HU values of T2 and T9 during preoperative chest CT interpretation. For patients with HU values lower than the recommended thresholds, additional evaluation with DEXA or early initiation of bone-targeted interventions may be considered, thus enabling timely identification and management of individuals at high risk for osteoporosis. Nevertheless, it should be emphasized that DEXA remains the gold standard for osteoporosis diagnosis and should continue to be prioritized in clinical decision-making [32].

Limitations

Our study has some limitations. First, the sex ratio of our recruited patients was approximately 6:1 (male: female), which may also affect the external validity of our findings. Previous studies have reported significant differences in bone mineral density loss and osteoporosis progression between men and women, particularly postmenopausal women. Therefore, future studies should include more balanced cohorts and perform analyses stratified by sex to further validate these results and enhance the clinical applicability and generalizability of their findings. In addition, our study was a single-center retrospective analysis with a limited sample size; larger sample sizes and multicenter prospective study designs are warranted to enhance the risk prediction for postoperative osteoporosis and vertebral fractures and to develop more robust diagnostic criteria for bone health management in lung transplant recipients. Furthermore, as differences in CT scanner models, scanning protocols, and image reconstruction algorithms across institutions may affect HU measurements and comparability of findings, we suggest that future multicenter collaborations focus on the standardization of scanning protocols, imaging parameters, and HU thresholds.

Despite these limitations, our study still demonstrates that routine preoperative chest CT can be an opportunistic screening tool for osteoporosis in lung transplant recipients. We hope that our findings help clinicians with early prediction of postoperative osteoporosis risk, thus reducing further bone loss after transplantation.

Conclusions

Pretransplant thoracic CT is a clinically feasible and opportunistic screening approach to stratify osteoporosis risk in lung transplant candidates while avoiding additional radiation exposure. By quantifying T2 and T9 vertebral HU values, clinicians can identify high-risk patients for timely intervention, potentially reducing fracture-related morbidity.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

AUC

Area under the receiver operating characteristic curve

BMD

Bone mineral density

BMI

Body mass index

CT

Computed tomography

COPD

Chronic obstructive pulmonary disease

DEXA

Dual-energy X-ray absorptiometry

FN/

Femoral neck

HU

Hounsfield unit

ROI

Region of interest

ROC

Receiver operating characteristic

T1–L1

All thoracic vertebrae and the first lumbar vertebra

LS

Lumbar spine

TH

Total left hip

TSH

Thyroid-stimulating hormone

Author contributions

F. Z. and Y.W.: Participated in study design, data analysis, and manuscript drafting. P.G.: Involved in data analysis, manuscript drafting, and translation. M. L. and X. W.: Contributed to study design, manuscript review, and quality control.

Funding

The study was supported by the health science and technology program of Zhejiang province (no. 2023KY770).

Data availability

No datasets were generated or analysed during the current study.

Declarations

Informed consent

The requirement for individual consent for this retrospective analysis was waived.

Conflict of interest

All authors declare that they have no conflict of interest.