Answers are grounded in this paper and cite their sources.
Introduction Immunotherapy targeting PD-1/PD-L1 has revolutionized NSCLC treatment but carries a risk of pneumonitis, a potentially severe immune-related adverse event. This study systematically evaluates pneumonitis incidence across PD-1/PD-L1 inhibitors and treatment lines. Methods A PRISMA-compliant review (PROSPERO: CRD420251026124) analyzed four databases through April 2025 for trials reporting pneumonitis in NSCLC patients receiving PD-1/PD-L1 inhibitors. Data extraction and risk-of-bias assessment (RoB-2 tool) were performed independently. A random-effects meta-analysis (DerSimonian-Laird method) assessed heterogeneity, sensitivity, and publication bias. Results Among 13,584 records, 85 studies (20,883 patients) were included. Most evaluated PD-1 inhibitors (nivolumab, pembrolizumab) or PD-L1 inhibitors (atezolizumab, durvalumab). Risk of bias was high in 50% of studies. Pooled all-grade pneumonitis incidence was 6% (95% CI: 5%–8%) for PD-1 inhibitors vs. 8% (95% CI: 4%–16%) for PD-L1 inhibitors, with no significant difference by treatment line. Grade ≥3 pneumonitis occurred in 3% (95% CI: 2%–3%) of patients on PD-1 inhibitors and 2% (95% CI: 1%–2%) on PD-L1 inhibitors, with significant variation in PD-L1 subgroups by line [first-line: 2% (95% CI: 1%–2%) vs. second-line: 1% (95% CI: 0%–1%); P=0.0078]. Grade 1–2 pneumonitis ranged from 1%–5%, grade 3–4 from 1%–3%, and fatal (grade 5) events were <1% (95% CI: 0%–2%) overall, though incidence varied by agent. Heterogeneity was higher in PD-L1 analyses (I² = 97.2% for all-grade events). Conclusion PD-1/PD-L1 inhibitors pose a low but clinically relevant pneumonitis risk in NSCLC, with PD-L1 agents showing marginally higher all-grade incidence. Severe cases are rare, particularly with second-line PD-L1 use. These findings underscore the need for vigilant monitoring, especially in high-risk patients. Mechanistic studies and improved risk stratification tools are warranted. Systematic Review Registration https://www.crd.york.ac.uk/PROSPERO/view/CRD420251026124 , identifier CRD420251026124.
No takes yet. Share an insight, caveat, or question.
Hammadeh et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: