Page 75 - ADULT-ONSET ASTHMA PREDICTORS OF CLINICAL COURSE AND SEVERITY

DIAGNOSTIC ACCURACY OF MINIMALLY INVASIVE MARKERS FOR DETECTION OF AIRWAY EOSINOPHILIA IN ASTHMA
methods for pa ent sampling (n=22; 69%) and/or blinding of the index test (n=20; 63%) and/ or blinding of the reference standard (n=18; 56%) were o en unclear.
META-ANALYSIS: AUCS
All extracted and obtained diagnos c accuracy data for markers and reference standards are summarized in Appendix 5. Results of meta-analyses of AUC es mates are presented in Table 1, with detailed results in Appendix 6.
Adults
Five di erent de ni ons of airway eosinophilia had been used across studies, most o en based on sputum eosinophils ≥2% or ≥3%. The prevalence of eosinophilia ranged from 20% to 88%.
We obtained diagnos c accuracy data for nine markers, but only FeNO, blood eosinophils, total Immunoglobulin E (IgE), serum perios n, serum Eosinophil Ca onic Protein and Exhaled Breath Condensate pH had been inves gated in more than one study (Table 1). When we pooled data, independent of which reference standard or airway eosinophilia de ni on had been used, the summary AUC of these markers never exceeded 0.78. We found substan al heterogeneity in most analyses (Appendix 6).
FeNO (17 studies; 3,216 pa ents), blood eosinophils (14 studies; 2,405 pa ents) and IgE (7 studies; 942 pa ents) have been inves gated in more than two studies, with pooled AUC es mates of 0.75 (range 0.59-0.88), 0.78 (0.63-0.91) and 0.65 (0.56-0.69), respec vely. We repeated these meta-analyses for studies that had used sputum eosinophils ≥3% and eosinophils ≥2% as the de ni on of airway eosinophilia (Appendix 7), but the summary AUCs were barely a ected: 0.74 (range 0.52-0.88) and 0.73 (0.53-0.81), respec vely, for FeNO; 0.78 (0.63-0.91) and 0.78 (0.66-0.84) for blood eosinophils; and 0.63 (0.56-0.69) and 0.66 (0.61-0.68) for IgE.
Perios n showed promising performance in one study (AUC 0.84), but these results were not replicated in a second study (AUC 0.55)14. Nasal lavage eosinophils (AUC 0.88) and a model based on exhaled Vola le Organic Compounds (VOCs; AUC 0.98) showed high accuracy, but were only inves gated in single studies.
Three studies reported combina ons of markers, but none of these showed a signi cant improvement in the diagnos c accuracy compared to single markers (data not shown).
Comparisons between published and unpublished diagnos c accuracy data for FeNO, blood eosinophils and IgE are provided in Appendix 8. Adding unpublished data led to a considerable increase in precision, but did not a ect summary es mates of accuracy.
Children
Five di erent de ni ons of airway eosinophilia had been used across studies, most o en based on sputum eosinophils ≥2.5% (Appendix 5). The prevalence of eosinophilia ranged
73