Longitudinal changes in skeletal muscle index and their prognostic significance in patients with advanced lung cancer: a retrospective cohort study

Technical Summary: Longitudinal Changes in Skeletal Muscle Index and Prognostic Significance in Advanced Lung Cancer

Problem Statement
Sarcopenia, defined as the progressive loss of skeletal muscle mass and function, is a recognized prognostic factor in advanced lung cancer. However, existing literature predominantly relies on single time-point measurements (baseline), implicitly assuming a linear decline in muscle mass. This static approach fails to capture the dynamic nature of body composition, which is influenced by tumor burden, systemic inflammation, nutritional status, and treatment toxicity. Consequently, the prognostic value of longitudinal changes in the Skeletal Muscle Index (SMI) remains poorly characterized, and the temporal patterns of muscle decline during treatment are not well understood.

Methodology
This retrospective cohort study analyzed 316 patients with newly diagnosed advanced lung cancer treated at Nanfang Hospital between January 2022 and December 2024.

• Data Acquisition: SMI was quantified using serial chest CT scans at the T12 vertebral level at four time points: baseline (T₀), first follow-up (T₁, ~3 months), second follow-up (T₂, ~6 months), and third follow-up (T₃, ~12 months).
• Measurement Protocol: A deep learning-based automated segmentation tool (TotalSegmentator v2.0) was used to segment skeletal muscle, with area measurements performed in 3D Slicer. Muscle tissue was identified using Hounsfield unit thresholds (−29 to +150 HU). SMI was calculated as total skeletal muscle area (cm²) divided by height squared (m²).
• Variable Definition: The rate of SMI change was calculated as the percentage change per month. Changes were categorized into early (T₀ to T₁), mid-term (T₁ to T₂), late (T₂ to T₃), and total phases.
• Statistical Analysis: Progression-free survival (PFS) was the primary endpoint. Optimal cut-off values for SMI change rates were determined using maximally selected rank statistics. Associations were evaluated using Kaplan–Meier curves and Cox proportional hazards models, adjusted for covariates including age, sex, BMI, tumor stage, brain metastasis, and treatment modality. Model performance was assessed via Harrell's C-index and Akaike information criterion (AIC).

Key Results

• Baseline vs. Longitudinal: While higher baseline SMI was significantly associated with improved PFS (HR = 0.895, P < 0.001), this prognostic effect attenuated over time. Among patients surviving beyond one year, baseline SMI was no longer a significant predictor (P = 0.682).
• Trajectory Pattern: The study observed a consistent non-linear "fast–slow–fast" trajectory of SMI decline across the cohort. SMI declined in 98.1% of patients.
• Prognostic Power of Change: Longitudinal SMI change emerged as a robust predictor of PFS throughout the follow-up period. The early change rate (baseline to T₁) demonstrated the strongest association with disease progression (HR = 0.849, P < 0.001).
• Optimal Cut-off: The optimal cut-off for monthly SMI decline was determined to be 0.89%. Patients with a monthly decline exceeding this threshold had a significantly worse one-year PFS (52.5%) compared to those with less decline (74.5%). Each 1% per month reduction in the rate of SMI decline (slower muscle loss) was associated with a 15.1% lower risk of disease progression.
• Model Performance: A combined model incorporating both baseline SMI and longitudinal SMI change showed improved model fit (lower AIC) compared to a model using baseline SMI alone, although the C-index showed a slight decrease, suggesting the added value lies in model calibration rather than pure ranking discrimination.
• Subgroup Findings: The prognostic value of baseline SMI was most pronounced in patients receiving targeted therapy (HR = 0.857), whereas associations in chemotherapy and immunotherapy subgroups were not statistically significant, likely due to smaller sample sizes.

Key Contributions

1. Characterization of Trajectories: The study provides empirical evidence of a non-linear "fast–slow–fast" decline pattern in skeletal muscle mass during treatment for advanced lung cancer, challenging the assumption of linear decline.
2. Dynamic vs. Static Prognosis: It demonstrates that while baseline SMI reflects initial physiological reserve relevant to early outcomes, longitudinal SMI changes capture ongoing disease-related catabolic processes and provide sustained prognostic information beyond one year.
3. Clinical Thresholds: The identification of a specific optimal cut-off (0.89% monthly decline) offers a quantitative metric for risk stratification that can be applied during routine clinical monitoring.
4. Methodological Feasibility: The study validates the use of T12-level measurements from routine chest CTs as a practical alternative to lumbar (L3) assessments, enabling repeated longitudinal monitoring without additional radiation exposure or cost.

Significance and Claims
The authors claim that longitudinal decline in skeletal muscle mass provides clinically meaningful prognostic information that extends beyond baseline measurements. The study suggests that serial SMI assessment serves as a useful tool for risk stratification and disease monitoring, particularly for identifying patients with rapid muscle loss who may benefit from early supportive interventions.

The paper posits that a dual approach to risk assessment is necessary: using baseline SMI for initial risk stratification and dynamic SMI monitoring for ongoing evaluation. The findings support the integration of routine CT-based body composition monitoring into the longitudinal management of advanced lung cancer patients. The authors explicitly state that their findings provide a rationale for future interventional studies targeting patients identified as high-risk based on early SMI decline, though they acknowledge the need for prospective validation and caution regarding limitations such as the retrospective design and lack of functional sarcopenia measures.