Pre-onset increases in neutrophil-to-lymphocyte ratio as an identifier of mild and severe immune checkpoint inhibitor–related pneumonitis in lung cancer.

Abstract

e24188 Background: Immune checkpoint inhibitor (ICI)–related pneumonitis is an uncommon but potentially life-threatening toxicity. Reliable biomarkers that predict pneumonitis prior to clinical onset are lacking. We evaluated whether dynamic changes in the neutrophil-to-lymphocyte ratio (NLR) preceding pneumonitis onset are associated with severity. Methods: Lung cancer patients treated with ICI-based therapy between October 2014 and December 2025 were retrospectively identified from an institutional pharmacy database. Pneumonitis events were graded per CTCAE v6.0 and categorized as mild (grade 1–2) or severe (grade ≥3). Serial complete blood counts were used to calculate baseline NLR and longitudinal pre-onset NLR metrics, including maximum NLR, absolute and percent change, and NLR slopes over 30- and 60-day intervals. Analyses were conducted at the pneumonitis-event level. Group comparisons used Wilcoxon rank-sum tests. Discriminatory performance was assessed using receiver operating characteristic (ROC) analysis, and associations with severe pneumonitis were evaluated using logistic regression. An optimal NLR slope threshold was identified by ROC analysis to define pre-event inflammatory acceleration. Results: Among 450 ICI-treated patients, 31 patients experienced 38 pneumonitis events (21 mild, 17 severe). Tumor types: adenocarcinoma/squamous cell carcinoma/small cell lung cancer/other malignancies 55%/26%/13%/6%. Median age at diagnosis: 72 years (IQR 65–76). Male: 55%. Race/ethnicity: Non-Hispanic White/non-White 77%/23%. Baseline NLR did not differ between mild and severe pneumonitis (median 3.99 vs 4.76; p = 0.21). In contrast, severe pneumonitis was associated with higher maximum NLR, greater absolute and percent increases in NLR, and steeper pre-event NLR slopes (all p < 0.01). The 30-day pre-event NLR slope was significantly higher in severe compared with mild pneumonitis (median 0.22 vs 0.02 units/day; p = 0.001). ROC analysis demonstrated good discrimination for severe pneumonitis (AUC 0.80) and identified a slope threshold of 0.04 units/day, which detected pre-event NLR acceleration in 94.1% of severe pneumonitis with a median of 15.5 days (IQR 37.3) and in 52.4% of mild pneumonitis with a median of 16.0 days (IQR 28.5) prior to clinical diagnosis. Each 0.1-unit/day increase in the 30-day NLR slope was associated with increased odds of severe pneumonitis (OR 1.68, 95% CI 1.15–3.18). Conclusions: Short-term pre-onset increases in NLR, particularly 30-day slopes, identify patients at increased risk for severe ICI-related pneumonitis. Longitudinal NLR monitoring may support earlier risk stratification and proactive clinical management, warranting prospective validation.