Findings in Relation to What Is Known

The results of this systematic review are consistent with other systematic reviews that addressed diagnostic performance of FeNO testing (KQ 1.a) and clinical utility of FeNO measurements to select medication option (KQ 1.c); whereas to our knowledge, no systematic reviews have addressed clinical utility of FeNO measurements in monitoring disease activity and asthma outcomes (KQ 1.b), clinical utility of FeNO measurements to monitor response to treatment (KQ 1.d), and FeNO testing in predicting the future development of asthma (KQ 1.e). In terms of diagnostic accuracy, Li et al. reported pooled estimates of sensitivity, specificity, and DOR of 0.78, 0.74 and 11.4.¹⁹¹ Tang et al. evaluated the diagnosis of asthma in children and reported pooled estimates of sensitivity, specificity, and DOR of 0.79, 0.81 and 16.5.¹⁹² Guo et al reported pooled estimates of sensitivity, specificity, and DOR of 0.72, 0.78 and 15.9.¹⁹³ The highest DOR (i.e. diagnostic accuracy) was observed in steroid-naive and nonsmoking patients.¹⁹³ In terms of tailoring asthma management using FeNO based algorithms, two recent Cochrane systematic reviews reported that these strategies reduced exacerbations in strategies for adults and children without a significant impact on other outcomes.^{194, 195} Although not outcomes of interest in our systematic review, total ICS dose and final mean FeNO level were also not statistically different between the FeNO-based approach and standard management.^{194, 195}

Limitations

For several of the key questions (KQ 1.b–e), studies were quite heterogeneous in terms of design, population, control tests, control strategies, and outcome measures; which led to narrative evidence synthesis and narrative rating of the strength of evidence. Narrative evidence synthesis is helpful for decision making; however, it does not provide a single best estimate; which is a limitation. Studies were overall small despite the fact that asthma is a very common condition. We also found limited data on baseline severity and large variations in FeNO protocols, which makes interpretation of the body of evidence challenging.

For the diagnostic accuracy question (KQ 1.a), there were several limitations. One challenge relates to the fact that there is no true gold standard of diagnosing asthma. Although we did not rate label studies as having increased risk of bias because of this issue, we recognize that it can impact diagnostic accuracy. In addition, a wide range of reference tests were reported. We categorized these reference tests as clinical diagnosis, positive bronchial challenge test, or a combination of clinical diagnosis, positive bronchial challenge, and/or bronchodilator response. However, significant heterogeneity still exist, such as to how and when these tests were administered. The studies reported a wide range of cutoffs from 0.8 ppb to 85 ppb. Although categorizations of <20, 20–30, 30–40 and >=40 ppb helped reduce heterogeneity and facilitated meta-analyses, we were not able to definitively present a best cutoff overall or within each category. We were also not able to conduct some planned subgroup analyses because of lack of data, including asthma phenotype, adequate testing procedures, body mass index (BMI) or weight, manufacturer and device model, and exhalation flow rate.

Applicability

The age of participants in the studies did not commonly conform to the definitions used in National Heart, Lung and Blood Institute prior asthma guideline (i.e. adults defined as 12 years of age or older)¹. Therefore, applicability may be affected when guideline developers provide recommendations using diiferent age cutoffs. Otherwise, most studies reported on patients with asthma commonly seen in practice. FeNO measurements in the included studies were for the most part consistent with the American Thoracic Society/European Respiratory Society 2005 guidelines¹⁹⁶ on the measurement of lower respiratory nitric oxide with the standard flow rate of 0.05L/second (body temperature [37° C] and pressure, saturated). The majority of studies did not include specific data on potential confounders including diet, use of mouthwash, and possible respiratory tract infections at the time of measurement. Such information is important for those developing institutional protocols for FeNO testing.

Clinicians considering FeNO as an adjunct to diagnose asthma should expect a fair number of false negatives (that is larger with higher test cutoffs) and an even a larger number of false positives (that is larger with lower test cutoff). The prevalence of asthma in the population being tested also impacts the expected positive and negative predictive values. Using several plausible asthma prevalence values in Figures 13 and 14, we simulate the number of false negative and false positive results expected in 1,000 patients tested for asthma using various FeNO test cutoffs. As the FeNO test sensitivity goes up (i.e. lower cutoff) the percentage of false negatives goes down, but the percentage of false positives goes up. Additionally as the prevalence of asthma increases in the screened population, the positive predictive value for confirmed asthma also increases.

Figure 13

False negatives per 1,000 patients.

Figure 14

False positive per 1,000 patients.

Suggestions for Future Research

Studies with better stratification according to asthma phenotype are needed (eosinophilic/versus non–eosinophilic) to identify populations who may benefit from serial FeNO measurement. Blood eosinophilia and atopy are likely good surrogates for airway eosinophilia and can be used to aid stratification of patients enrolled in studies. The field also needs studies of FeNO-based adherence monitoring programs that specifically evaluate cost effectiveness, acceptability, feasibility, and outcomes of such programs. These studies should also be either group stratified as above, or focus on atopic or eosinophilic patients.

In this review, we demonstrated that FeNO can identify those who will be steroid responsive; therefore, studies of FeNO-based medication titration are needed and should focus on symptomatic patients with previously documented elevated FeNO. Studies evaluating disease activity and outcome, should use validated measures of activity and well defined outcomes.

The role of serial FeNO measurements in children ages 0–5 year who develop illness associated with wheezing remains unclear. Cohort studies of such infants with follow up into later years of childhood and adolescence are needed to establish if persistently elevated levels correlate with increased risk of ultimate asthma diagnosis. This question is of particular importance, because the best biomarker we have at this time to predict asthma in this setting is the presence of eczema, which can be subjective. In addition, some children (regardless of age) often suffer from wheezy bronchitis, also known as wheezing associated respiratory infections. These are discrete illnesses with good prognosis that are quite common in pre-school age. Despite the benign outcome, many of these children still receive oral steroids. Would point of care FeNO measurements identify the children who do not require oral steroids? Such knowledge might address a very common clinical problem and spare children and their parents the adverse effects of steroids.

This review has yeilded a very small body of evidence on geriatric asthma. It will be important to determine the clinical utility of FeNO in a population that was underrepresented in the current literature.

Future research should also address the effect of emerging treatments such as anti-IL13 and anti-IL5 drugs on FeNO levels. Knowledge of such effect may demonstrate a role of FeNO in monitoring the use and adherence to some of these treatments and not others.

A challenge we faced in this review is to define the reference test for asthma diagnosis. Future studies should be explicit in describing the reference standard and use the modern testing approach recommended in current clinical practice guidelines; which may improve accuracy of diagnosis and make evidence more relevant. Similarily, studies should attempt to be consistent with guideline recommendations in definition of variables such as age, FeNO protocols and cutoffs, and asthma control categories, to further enhance applicability. Studies should also investigate factors that may affect FeNO diagnostic accuracy, including asthma phenotype, adequate testing procedures, body mass index (BMI) or weight, manufacturer and device model, and exhalation flow. More research should be done to evaluate diagnostic utility of FeNO testing in picking up asthma in a general population not on any type of treatment.