Journal article
CT Emphysema and Deep Learning-derived Vertebral Bone Loss in Individuals without COPD: Findings from MESA
Radiology, Vol.319(1), e252251
04/2026
DOI: 10.1148/radiol.252251
PMID: 42012343
Abstract
Background It is unknown whether subclinical emphysema-like changes, quantified at chest CT, are associated with loss of bone mineral density (BMD) in individuals without clinical chronic obstructive pulmonary disease (COPD). Purpose To identify early imaging markers of lung and bone health using deep learning-based imaging analysis in a large multiethnic cohort. Materials and Methods Chest CT scans obtained using the SubPopulations and InteRmediate Outcome Measures in COPD Study (SPIROMICS) protocol during examination 5 and 6 of the prospective Multi-Ethnic Study of Atherosclerosis (MESA) were secondarily analyzed (April 2010-March 2018). Percentage emphysema was quantified at examination, and thoracic vertebral BMD was assessed at both examinations using a validated deep learning-based segmentation model. Participants with clinical COPD as determined on the basis of spirometry or self-reports were excluded. Linear mixed-effects models were constructed and adjusted for demographic characteristics and covariates, including smoking status, physical activity, and scanner type, and the interaction effect of sex was analyzed. Results The cross-sectional analysis included 2312 participants (median age, 67 years [IQR, 61-75 years]; 1285 female participants), and the longitudinal study included 1109 participants (median age, 65 years [IQR, 60-72 years]; 614 female participants) with available follow-up CT data. A greater percentage of emphysema-like changes was associated with lower BMD (β = -1.14 mg/cm
[95% CI: -1.76, -0.53]) cross-sectionally and with greater annual BMD loss (β = -0.07 mg/cm
per year [95% CI: -0.13, -0.01]) longitudinally. Interaction effects were identified for sex (
< .001) and race or ethnicity (
= .049). In stratified models, the associations were significant for men (β = -0.38 mg/cm
per year [95% CI: -0.48, -0.28]) and Black/African American participants (β = -0.24 mg/cm
per year [95% CI: -0.39, -0.09]). The Johnson-Neyman method revealed more than 2.7% emphysema as a threshold for a decrease in BMD among men, which corresponded to 39% of the male participants. The results remained robust after adjustment for pulmonary function and chronic respiratory symptoms. Conclusion In individuals without COPD, a greater percentage of emphysema-like changes was independently associated with faster vertebral bone loss, particularly in men, suggesting that subclinical emphysema may be a novel imaging marker of systemic skeletal decline. Clinical trial registration no. NCT0000548 © RSNA, 2026
See also the editorial by Fukuda in this issue.
Details
- Title: Subtitle
- CT Emphysema and Deep Learning-derived Vertebral Bone Loss in Individuals without COPD: Findings from MESA
- Creators
- Elena Ghotbi - Johns Hopkins UniversityQuincy A Hathaway - University of PennsylvaniaPayam Jannatdoust - Tehran University of Medical SciencesRoham Hadidchi - Johns Hopkins UniversityMichael P Bancks - Wake Forest UniversityDavid A Bluemke - University of Wisconsin–MadisonMatthew Budoff - The Lundquist InstituteWendy Post - Johns Hopkins UniversityEric A Hoffman - University of IowaBenjamin M Smith - Columbia University Irving Medical CenterJoão A C Lima - Johns Hopkins UniversityR Graham Barr - Columbia University Irving Medical CenterShadpour Demehri - Johns Hopkins University
- Resource Type
- Journal article
- Publication Details
- Radiology, Vol.319(1), e252251
- DOI
- 10.1148/radiol.252251
- PMID
- 42012343
- ISSN
- 1527-1315
- eISSN
- 1527-1315
- Publisher
- RSNA
- Grant note
- R01 HL146666 / NHLBI NIH HHS N01HC95169 / NHLBI NIH HHS N01HC95161 / NHLBI NIH HHS 75N92020D00002 / NHLBI NIH HHS N01HC95164 / NHLBI NIH HHS 75N92020D00005 / NHLBI NIH HHS N01HC95167 / NHLBI NIH HHS N01HC95159 / NHLBI NIH HHS 75N92020D00003 / NHLBI NIH HHS N01HC95163 / NHLBI NIH HHS
- Language
- English
- Date published
- 04/2026
- Academic Unit
- Roy J. Carver Department of Biomedical Engineering; Radiology; Internal Medicine
- Record Identifier
- 9985157605102771
Metrics
1 Record Views