CT Body Composition Predicts Severe Viral Pneumonia Risk
Key Takeaways
- In hospitalized adults with viral pneumonia at a single center in China, lower T12 erector spinae volume and higher T12 visceral adipose volume were independently associated with severe disease.
- Severity-linked body composition differences were concentrated at T12, while most T4 and T8 measures were not significantly associated.
- Within the 65-patient severe subgroup, unadjusted comparisons showed lower subcutaneous fat across thoracic levels in the 24 patients with 30-day adverse outcomes, whereas visceral fat and muscle volume were not associated in those comparisons; those patients were also older and more often male, and the study did not present an adjusted adverse-outcome model.
- A combined model using oxygen therapy, CRP, T12-ESV, and T12-VAV discriminated severe viral pneumonia better than the clinical-only or imaging-only models.
Baseline chest CT body composition was measured at the T4, T8, and T12 costovertebral joint levels and at the corresponding whole-vertebral levels in 3D-Slicer version 5.2.1. Thresholds were −29 to 150 Hounsfield units (HU) for muscle and −150 to −30 HU for adipose tissue; whole-vertebral volumes were standardized by vertebral slice number, and single-slice measures were also normalized to height.
The cohort was predominantly elderly and male, and the clearest severity-linked body composition differences were concentrated at T12 rather than across all measured sites. Most T4 and T8 measures were not significantly associated with severe disease. After adjustment, lower T12 erector spinae volume (T12-ESV) was independently associated with severe viral pneumonia (OR 0.823, 95% CI 0.733–0.924), and higher T12 visceral adipose volume (T12-VAV) was independently associated with severe viral pneumonia (OR 1.030, 95% CI 1.014–1.047).
Within the 65-patient severe subgroup, unadjusted comparisons showed lower subcutaneous fat across thoracic levels in the 24 patients with 30-day adverse outcomes, whereas visceral fat and muscle volume were not associated in those comparisons. The study also reports that those patients were older and more often male, and it does not present an adjusted adverse-outcome model. The combined CT and clinical model for severe viral pneumonia using oxygen therapy, C-reactive protein (CRP), T12-ESV, and T12-VAV had an AUC of 0.843 versus 0.796 for the clinical model and 0.735 for the imaging model, with acceptable calibration and a supportive 5-fold internal validation signal.
These findings are observational and do not establish causality. The authors noted that the cohort was relatively small, drawn from a single center, and composed largely of older adults with substantial medical complexity. They also reported no external validation cohort, no comparison group with other pneumonia types, viral classification based on a six-virus panel, no rigorous paired agreement testing across quantification methods, and unresolved questions about missing-input robustness and overfitting.
The authors concluded that baseline chest CT body composition, especially T12 erector spinae and visceral adipose measures, was linked to viral pneumonia severity in hospitalized adults.
Clinician Questions
How was severe viral pneumonia defined in this CT body composition cohort?
Hospitalized adults with viral pneumonia were categorized using Torres et al. severe pneumonia criteria, and that framework separated 75 non-severe cases from 65 severe cases in this cohort.
Which chest CT levels showed the strongest body composition signal for viral pneumonia severity?
T12 showed the strongest body composition signal. Investigators examined single-slice, whole-vertebral, standardized, and height-normalized measures at T4, T8, and T12, but the clearest recurring severity-associated differences centered on T12 while most T4 and T8 measures were not significantly associated.