Nitric oxide uptake in the lungs in smokers with emphysema

Hypothesis
Since its introduction in 1983, pulmonary diffusing capacity for nitric oxide (DLNO) has been underutilized clinically, partly due to limited awareness of its potential alongside pulmonary diffusing capacity for carbon monoxide (DLCO). This study evaluates DLNO’s effectiveness in predicting and classifying emphysema compared to DLCO or spirometry alone using standardized z-scores. It was hypothesized that DLNO z-scores would outperform conventional diagnostic metrics, either alone or in combination with DLCO.

Data Overview
This analysis pooled data from four published studies: (1) Int J Chron Obstruct Pulm Dis (2024), (2) Pneumologie (2021), (3) Eur Respir J (1990), and (4) Respir Med (2009). Studies 2, 3, and 4, which used breath-hold times (BHT: 8–12 s), were analyzed across three European hospitals. Participants were categorized as smokers with CT-diagnosed emphysema or smokers without emphysema.

Data Processing
Quality control excluded cases with BHTs outside 8–12 s, alveolar volume/total lung capacity (VA/TLC) ≥ 1.00, inspired volume/forced vital capacity (IV/FVC) < 0.85, or forced expiratory volume in one second (FEV1)/FVC ≥ 1.00. Non-emphysema cases with a residual volume/total lung capacity (RV/TLC) ratio < 0.20 were also excluded.

Analysis
After filtering, 408 patients remained (20% emphysema, 80% no emphysema). LASSO regression identified key variables, including z-scores of FEV1/FVC, FEV1, DLCO, TLC, RV/TLC, DLNO, KNO, KCO, and VA. Binary logistic regression was then applied, and model performance was assessed using Matthews correlation coefficient (MCC) for classification and the area under the receiver operating characteristic curve (AUROC) for discrimination.

Notable Findings
The best model incorporated z-scores of DLNO, FEV1, and TLC, yielding the lowest Bayesian Information Criterion (BIC). Substituting DLCO for DLNO increased the BIC, indicating a worse fit. The MCC for this model was 0.80 (95% CI: 0.68–0.89), and AUROC was 0.97 (95% CI: 0.95–0.98). A summed z-score model of DLCO, FEV1/FVC, and FEV1 also performed well but was less effective than the DLNO-based model.

Implications
No single predictor provided optimal diagnostic accuracy. Instead, DLNO combined with FEV1 and TLC significantly improved classification, outperforming traditional spirometry-based models. These findings support integrating DLNO into routine emphysema diagnostics, as it surpasses DLCO in predictive value and enhances clinical assessments.