Accuracy of type 2 inflammatory biomarkers to diagnose asthma in adults and children: a meta-analysis of 30 studies with prospective validation in an adult primary care cohort

Le résumé fourni par la source

BACKGROUND: Asthma diagnosis is challenging, partly due to low yield of bronchodilator reversibility and limited access to bronchial provocation tests (BPTs). Fractional exhaled nitric oxide (FeNO) and blood eosinophil count (BEC) have been proposed as rule-in biomarkers. We aimed to evaluate the diagnostic accuracy of these biomarkers in adults and children, focusing on primary care applicability. METHODS: In this systematic review and meta-analysis, we searched MEDLINEand other databases from inception to June 4, 2024, for eligible diagnostic studies (randomised trials, cohort, cross-sectional, case-control, or case series). We included studies comparing FeNO or BEC with BPT or bronchodilator reversibility plus BPT in patients aged 5 years or older with suspected asthma and normal spirometry. Diagnostic performance in adults (aged ≥18 years) and children (aged 5-17 years) was estimated by the primary outcome measure of area under the receiver operating characteristic curve (AUC). Rule-in thresholds (prespecified) and rule-out thresholds (post hoc) were derived. In a single-centre cohort study (DIVE), analyses were prospectively replicated in bronchodilator reversibility-negative individuals referred for a BPT by primary care. Patients underwent FeNO and BEC immediately before a standardised methacholine BPT. The primary outcome for DIVE was biomarker performance for positive BPT, ascertained by AUC for FeNO, BEC, and the combination of FeNO and BEC. Post-hoc analyses addressed potential confounding by inhaled corticosteroid exposure. The systematic review and meta-analysis was registered with PROSPERO (CRD42023489738). The DIVE study was registered with ClinicalTrials.gov (NCT05592519). FINDINGS: 22 adult studies with 7141 patients and nine paediatric studies with 989 patients were included for meta-analysis (30 studies overall, with one study in both adults and children). FeNO diagnostic accuracy was good in adults (AUC 0·80 [95% CI 0·73-0·85]) and fair in children (0·75 [0·65-0·82]). In adults, FeNO of 48 ppb or higher achieved a positive likelihood ratio of 10·1 (95% CI 8·9-11·2). At a 20% prevalence reflecting primary care, a FeNO of 65 ppb or higher achieved a positive predictive value (PPV) of 80% (95% CI 72-86), and FeNO of less than 20 ppb achieved a negative predictive value of 93% (91-94) with 18% (12-25) missed diagnoses. In children, at the 56% aggregated study prevalence, FeNO of 30 ppb or higher achieved a PPV of 80% (95% CI 73-84). In adults, BEC showed weak stand-alone accuracy (AUC 0·65 [95% CI 0·59-0·71]) and added no value to FeNO. The DIVE cohort study (22 [18%] of 123 patients BPT positive) validated results of the meta-analysis in adults, with similar AUCs (0·77 [95% CI 0·65-0·89] for FeNO alone and 0·67 [0·53-0·80] for BEC alone). FeNO thresholds correctly classifyied rule-in decisions (FeNO >65 ppb: PPV 73% [95% CI 39-94]) and rule-out decisions (FeNO <20 ppb: NPV 92% [82-97]) with 89% overall accuracy. In both the meta-analysis and the DIVE cohort, FeNO diagnostic accuracy declined within 2 weeks of inhaled corticosteroid exposure. INTERPRETATION: In adults and children with suspected asthma, FeNO predicted airway hyper-responsiveness and enabled clinically applicable rule-in and rule-out thresholds. BEC performed poorly alone and added no value to FeNO. Findings and thresholds were replicated in a prospective adult primary care cohort; a similar study in children is ongoing. FeNO-guided diagnostic algorithms using context-specific rule-in and rule-out thresholds, while accounting for recent inhaled corticosteroid exposure, warrant consideration. FUNDING: Fonds de Recherche du Québec-Santé, Association Pulmonaire du Québec, Québec Ministry of Health and Social Services, AstraZeneca, Opération Enfant Soleil, and NIOX Group.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.