KEY POINTS
- The single-centre retrospective study included 231 patients receiving radical-intent radiotherapy for stage I–IV non-small cell lung cancer and analyzed 2,708 computed tomography scans over a median 22 months of follow-up. Treatment included stereotactic body radiotherapy in 132 patients and conventionally fractionated definitive radiotherapy in 99.
- Automated segmentation quantified skeletal muscle at L1 and left ventricular myocardial mass longitudinally. The muscle analysis included 184 patients and 1,647 scans, while the myocardial analysis included 150 patients and 1,665 scans, allowing patient-specific monthly rates of anatomical change rather than two-scan comparisons.
- Skeletal muscle loss faster than 0.4% per month identified a substantially higher-risk group: mortality was more than fourfold higher than in patients with preserved or minimally declining muscle (HR 4.41, 95% CI 2.46–7.91; p<0.005). The optimism-corrected concordance index was 0.70.
- Left ventricular myocardial change showed a U-shaped rather than linear association with survival. Atrophy faster than 0.3% per month was associated with HR 4.12 (95% CI 1.65–10.27), while hypertrophy greater than 0.3% per month was associated with HR 7.90 (95% CI 2.82–22.15); both had p<0.005, with a concordance index of 0.75.
- A conventional linear model failed to detect the myocardial-survival association, illustrating why the non-linear analysis mattered. Both substantial myocardial gain and loss, rather than simply declining myocardial mass, defined the adverse phenotype.
- Exploratory dose modelling associated esophageal V10 Gy with skeletal muscle loss (OR 1.40, 95% CI 1.04–1.89), aortic V5 Gy with myocardial atrophy (OR 1.71, 95% CI 1.08–2.70), and right-atrial V10 Gy with myocardial hypertrophy (OR 1.63, 95% CI 1.09–2.43).
- These dose models had only modest discrimination, with areas under the curve of 0.63–0.71, and the cohort required repeated post-treatment imaging. The findings therefore describe evolving prognostic signals rather than biomarkers currently capable of guiding radiotherapy planning before treatment.
CLINICAL TAKEAWAY
Routine surveillance imaging may contain useful information about evolving frailty and cardiac remodeling after thoracic radiotherapy. The striking survival associations justify prospective validation, but the study does not establish that radiation caused these changes or that modifying the identified dose metrics would improve outcomes.