Evaluating assumptions of definition-based pulmonary exacerbation endpoints in cystic fibrosis clinical trials

Highlights • Pulmonary exacerbation difference is an important clinical trial efficacy endpoint. • Regulators have stressed use of Fuchs criteria-based exacerbation definitions. • We compared criteria-based definitions to investigator assessments of 751 events. • Criteria-based definitions were insensitive, nonspecific, and prone to bias. • An alternative exacerbation definition is needed for future CF drug development.
Background Cystic fibrosis (CF) pulmonary exacerbations can be serious respiratory events and reduction in exacerbation rate or risk are important efficacy endpoints for CF therapeutic trials. Variability in exacerbation diagnoses and treatment have led drug developers to employ “objective” exacerbation definitions combining antimicrobial treatment (AT) and the presence of ≥4 of 12 respiratory criteria (first published by Fuchs et al. [NEJM 1994;331(10):637–42]). Assumptions underlying this approach have yet to be formally evaluated. Methods Respiratory events (RE) observed during a 48-week trial of ataluren (NCT02139306), a read-through agent for premature nonsense codons, were compared across six exacerbation definitions: any AT, intravenous AT (IVAT), ≥4 Fuchs criteria present, AT plus ≥4 Fuchs criteria, IVAT plus ≥4 Fuchs criteria, and investigator assessment. Fuchs definitions were evaluated by assessing missingness of individual criteria and associations between criteria presence and clinician exacerbation assessment. Results Among 751 RE, more than one third had ≥4 Fuchs criteria present but were not assessed as exacerbations by investigators. Data for ≥1 and for 4 Fuchs criteria, respectively, were missing for ~ 90% and >30% of RE. Only 6/12 Fuchs criteria were present more often when investigators assessed RE as exacerbations than when they did not. Conclusions “Objective” definitions have shortcomings inconsistent with their purpose of optimizing exacerbation capture in clinical trials : 1) they capture events clinicians do not consider exacerbations, 2) are prone to data missingness which can bias the likelihood of meeting the definition, and 3) employ criteria that are not associated with investigator assessment of exacerbation.