Page 84 - ADULT-ONSET ASTHMA PREDICTORS OF CLINICAL COURSE AND SEVERITY

ADULT-ONSET ASTHMA – PREDICTORS OF CLINICAL COURSE AND SEVERITY
data-driven selec on of an op mal cutpoint leads to over-op mis c es mates of sensi vity and speci city10. It could be more informa ve to report on sensi vity at a xed high speci city (e.g. 95%), or the other way around.
An ATS guideline on the Interpretation of FeNO for Clinical Applications strongly “recommends the use of FeNO in the diagnosis of eosinophilic airway in amma on”21. It also strongly “recommends that low FeNO less than 25ppb (20ppb in children) be used to indicate that eosinophilic in amma on and responsiveness to cor costeroids are less likely”, “that FeNO greater than 50ppb (35ppb in children) be used to indicate that eosinophilic in amma on and, in symptoma c pa ents, responsiveness to cor costeroids are likely” and “that FeNO values between 25ppb and 50ppb (20-35ppb in children) should be interpreted cau ously and with reference to the clinical context”. Our results are challenging this concept. Appendix 5 shows that, at FeNO thresholds below 25ppb, sensi vity ranges from 0.52 to 0.86 in adults. This means that of every 100 asthma c pa ents with airway eosinophilia tested by FeNO, up to 48 would be falsely considered as not having airway eosinophilia, and e ec ve treatment may be withheld from them. In children, sensi vity for FeNO thresholds below 20ppb ranges from 0.75 to 0.82, indica ng that up to 25 pa ents would be false nega ves. Although these thresholds may be relevant in speci c subgroups of asthma, these ndings indicate that FeNO results should be interpreted with much more cau on in the general asthma popula on than recommended by the ATS.
It is probably not surprising that the markers evaluated in our review generally were moderately accurate. The underlying biological mechanisms determining airway eosinophil counts are considerably di erent from those of some of the inves gated markers22. Several studies also showed signi cant variability in blood eosinophils23 and IgE24 in the same asthma c subjects in short periods of me. Some asthma c pa ents were shown to have persistently elevated FeNO levels, not suppressed by cor costeroid treatment and not re ec ng raised sputum eosinophils25. Cor costeroid treatment signi cantly in uences FeNO, blood eosinophils, IgE and sputum eosinophils26, but the rela ve magnitude of this e ect could vary across markers. Diagnos c accuracy may therefore be in uenced by treatment status. Also many other factors, such as age, gender, re ux disease, smoking, and atopy, have been shown to in uence FeNO levels27. This may also be the case with other markers and further compromises the iden ca on of an accurate minimally invasive test for airway eosinophilia.
Similar reproducibility problems could apply to the reference standard and target condi on. Although some studies showed that a threshold of 3% for sputum eosinophils is reproducible over me28, others found the phenotypic classi ca on of asthma to change frequently, both spontaneously and in response to treatment29. Longitudinal studies examining sputum cell counts in successive exacerba ons found considerable heterogeneity in the type of in amma on within the same individuals30. Consequently, a diagnosis of eosinophilic asthma based on a single sputum sample may be ques onable.
Based on our ndings, we discourage the use of FeNO, blood eosinophils, or IgE as single
82