Page 83 - ADULT-ONSET ASTHMA PREDICTORS OF CLINICAL COURSE AND SEVERITY

DIAGNOSTIC ACCURACY OF MINIMALLY INVASIVE MARKERS FOR DETECTION OF AIRWAY EOSINOPHILIA IN ASTHMA
COPD Overlap Syndrome) can be problema c. The accuracy of markers may vary across these di erent subgroups. The prevalence of airway eosinophilia also di ered considerably across studies. Diagnos c accuracy typically varies with clinical se ng, context, and prevalence. Although the results from the individual studies re ect considerable heterogeneity, we felt safe to draw conclusions because AUCs for FeNO, blood eosinophils and IgE consistently re ected moderate accuracy.
Combining markers with other clinical features in a predic on model is likely to improve diagnos c accuracy compared to single markers, but this has not been su ciently inves gated yet. All but three studies only reported on accuracy es mates of single markers. Since we did not have individual pa ent data (IPD), we were unable to further analyse the incremental value of combining markers.
The most robust evidence for the clinical value of detec ng airway eosinophilia comes from a Cochrane review that demonstrated that the frequency of asthma exacerba ons can be signi cantly reduced when tailoring inhaled cor costeroids on sputum eosinophilia3. For a marker to be able to replace induced sputum in this context, sensi vity, speci city and AUC should probably be at least above 90%, so that at most 10% of all pa ents will be misclassi ed and, poten ally, subjected to inappropriate clinical decisions. Our review shows that there are currently no single markers available with a large enough documented accuracy to ful ll these criteria. It must be noted, though, that recent guidelines recommending the use of sputum eosinophil counts in severe asthma, acknowledge that the quality of evidence is “very low”1;4. In addi on, they do not recommend sputum-guided treatment in the general asthma popula on. Some of the markers evaluated in this review on their own may have be er poten al in managing asthma than sputum eosinophil counts. This is illustrated by a recent study in which VOC-analysis predicted cor costeroid responsiveness with greater accuracy than sputum eosinophils19, and by another study which showed good response to Mepolizumab among pa ents with severe eosinophilic asthma as determined by blood eosinophils20. The la er study illustrates the accumula ng evidence for the poten al role of blood eosinophils as a predictor of responsiveness to novel targeted therapies against eosinophilic airway in amma on7.
Moderate accuracy does not necessarily make the inves gated markers useless. Markers can also be applied in a triage se ng, for example, for ruling-out (high sensi vity required) or ruling-in (high speci city required) airway eosinophilia. In case of a high speci city, those with a posi ve test result would be considered as eosinophilic. With a limited sensi vity, those with a nega ve test result would need to undergo further tes ng (e.g. sputum induc on). Most included studies only reported on the “op mal cutpoint” between sensi vity and speci city, based on the Youden index. When a marker is not su ciently accurate to replace the exis ng test, this op mal cutpoint is clinically not very prac cal because both sensi vity and speci city are typically subop mal at this cutpoint. Therefore, it does not inform the reader about the ability of the marker to rule-in or rule-out airway eosinophilia. Furthermore,
81