Real-World Lung Function and Symptom Response to Dupilumab in Type 2-Inflamed Chronic Obstructive Pulmonary Disease Across the Full Spectrum of Airflow Limitation: An Italian Multicentre Retrospective Cohort Study

Design and setting. Retrospective, observational, multicentre cohort study conducted at six Italian respiratory and internal medicine centres: San Carlo di Nancy Hospital (GVM Care & Research, Rome), Fondazione Policlinico Universitario Campus Bio-Medico (Rome), Azienda Ospedaliero-Universitaria Sant'Andrea (Sapienza University of Rome), Ospedale Cristo Re (Rome), Azienda Ospedaliera San Camillo-Forlanini (Rome) and Ospedale Isola Tiberina - Gemelli Isola (Rome). The study protocol (code PN002) was approved by the Comitato Etico Regionale Umbria on 20 May 2026. The study was conducted in accordance with the Declaration of Helsinki and is reported following the STROBE recommendations for observational research. Rationale. The BOREAS and NOTUS phase 3 trials established the efficacy of dupilumab in COPD with type 2 inflammation but restricted enrolment to a post-bronchodilator FEV1 between 30 and 70 percent predicted. Whether the functional response extends to patients outside that window has not been examined in a multicentre real-world cohort, and no published observational series has reported the response at the 12-week timepoint pre-specified in the registration trials.

Participants and treatment. Consecutive adults with a physician diagnosis of COPD who commenced dupilumab between October 2024 and April 2026 were considered. All received dupilumab 300 mg subcutaneously every two weeks as add-on therapy through the manufacturer's named-patient compassionate-use programme; background inhaled therapy was continued unchanged. Because compassionate-use access is typically requested for refractory disease and applies no ceiling on disease severity, the cohort is expected to be weighted towards the severe end of the type 2-inflamed COPD spectrum and should not be read as a representative sample of all candidates for biologic therapy.

Measurements. Spirometry was performed at each centre according to the American Thoracic Society and European Respiratory Society technical standards. All reported spirometric values are post-bronchodilator. Percent predicted values, z-scores and lower limits of normal were computed from the Global Lung Function Initiative 2012 equations for the Caucasian population. Airflow limitation was graded on post-bronchodilator FEV1 percent predicted using the Global Initiative for Chronic Obstructive Lung Disease thresholds. Symptom burden was quantified with the CAT and the mMRC dyspnoea scale. FeNO was measured before spirometry where locally available. Blood eosinophil counts were obtained from routine laboratory testing at the screening visit. Assessments were scheduled at baseline and at approximately 12 weeks, following local standard of care.

Analysis. Continuous variables are summarised as mean with standard deviation and as median with interquartile range; categorical variables as counts and percentages. Paired comparisons between baseline and 12 weeks were performed with the Wilcoxon signed-rank test, with the paired t-test reported alongside; 95 percent confidence intervals for mean differences were derived from the t distribution and standardised effect sizes are expressed as Cohen's dz. Two pre-specified analyses addressed generalisability and robustness: participants were classified as trial-eligible or trial-ineligible according to whether baseline post-bronchodilator FEV1 fell within the 30 to 70 percent predicted window, with between-group comparison by the Mann-Whitney U test; and heterogeneity across recruiting centres was assessed with the Kruskal-Wallis test and with a leave-one-centre-out sensitivity analysis. No imputation was performed for missing data; analyses used all available paired observations and the denominator is reported for every estimate.

Overlap with previous work. Twelve participants from one contributing centre were previously reported in a single-centre descriptive compassionate-use case series with a different analysis timepoint and a different set of endpoints. They are included here within the larger multicentre cohort, which addresses a distinct research question with a pre-specified primary variable.