Assessment of adherence to corticosteroids in asthma by drug monitoring or fractional exhaled nitric oxide: A literature review

Abstract Background Although the efficacy of corticosteroid treatment in controlling asthma is widely accepted, its effectiveness is undermined by poor adherence. However, the optimal method for measuring adherence to asthma therapy remains unclear. Objective To perform a review of the literature reporting biological, objective methods for assessing adherence to inhaled or oral corticosteroids in asthma; we included studies reporting direct measurement of exogenous corticosteroids in blood, or the effect of adherence on exhaled nitric oxide. Design We searched three databases MEDLINE (using both PubMed and Ovid), the Cumulative Index of Nursing and Allied Health Literature (CINAHL) and Web of Science for articles published between January 1975 and July 2020. Quality of the studies was assessed using the National Institute of Health checklist. Results From 2850 screened articles, 26 fulfilled the inclusion criteria. Measurement of blood prednisolone with or without cortisol was used in eight studies as a measure of oral corticosteroid adherence, and fractional exhaled nitric oxide (FeNO) in 17 studies for inhaled corticosteroid adherence. Inhaled corticosteroids were measured directly in the blood in one study. By direct measurement of drug levels in the blood, adherence rates to oral corticosteroids ranged from 47% to 92%, although the performance and timing of the test were often not known, making interpretation of findings and serum prednisolone concentrations difficult. FeNO is generally lower in adherent than non‐adherent patients, but no absolute cut‐off can be proposed based on the available data. However, a fall in FeNO following supervised inhaled corticosteroid dosing could indicate previous poor adherence. Conclusions and Clinical Relevance Despite prednisolone and cortisol levels commonly being used as adherence markers in clinical practice, further work is required to assess the influence of the dose and timing on blood levels. The promising findings of FeNO suppression testing should be explored in further prospective studies.


| INTRODUC TI ON
Asthma is the most common chronic inflammatory airway disease, affecting more than 300 million people worldwide. 1 Inhaled corticosteroids (ICS) are the foundation of daily asthma control medication, and in severe asthma are needed at high doses, often in combination with daily oral corticosteroids (OCS). 2 A high prevalence of poor adherence has been identified in all asthma severities, which impacts exacerbation and hospitalization rates, quality of life and pulmonary function. The impact on healthcare costs is complex, with non-adherence resulting in lower direct spending on inhalers, but potentially higher costs associated with exacerbations, systemic steroid toxicity and escalation to biological therapy. 1,[3][4][5][6] Measuring adherence to asthma medications is recommended in national and international guidelines, 2,7 but represents a significant challenge. Methods of monitoring adherence should be accurate, easy-to-use and cost-effective. They can be classified into objective and subjective methods, but all have drawbacks and none would be considered as a "gold standard." The most common subjective methods of adherence include self-report questionnaires, patient diaries, and assessment by healthcare providers. This is known to be unreliable, as patients particularly tend to overestimate their adherence. 3 Objective assessments, which include pill counting, prescription pick-ups and electronic device monitoring (smart inhalers), may also overestimate adherence where patients pick up or actuate their inhalers without inhaling the drug as prescribed. The effect of corticosteroid medications (pharmacodynamics) on blood biomarkers such as blood eosinophils [8][9][10] or cortisol 11 may be useful in ascertaining efficacy and toxicity, but this may not directly reflect adherence. The fractional exhaled nitric oxide test (FeNO) is an alternative biological non-invasive method used as a surrogate marker of ICS adherence, [12][13][14] due to the effect of ICS on nitric oxide synthase. 15 Direct drug monitoring in biological fluids is probably considered the most reliable and accurate method, as it at least provides evidence that the patient has taken the medication sometime in the recent past, according to the specific assay and pharmacokinetics.
Numerous studies have used blood assays to identify non-adherence in asthma patients regarding prescribed daily OCS by measuring prednisolone and cortisol levels, 9,16,17 and more recently, a similar approach has been reported for ICS. 18 We have conducted a review aiming to identify and summarize the available literature on two objective biological methods that have been used to assess the adherence to either ICS or OCS in adults and children with asthma: (1) detection of inhaled or oral corticosteroids in body fluid, and (2) level of exhaled nitric oxide. Due to the heterogeneity of the studies within each adherence method included, a meta-analysis was not conducted.

| Search strategy
This review was reported according to PRISMA (the Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines. 19 Although this was undertaken according to the standards of a systematic review in terms of search strategy, quality assessments and data extraction, PICO (Population, Intervention, Comparison and Outcome) reporting criteria were not suitable, due to multiple interventions of interest, and the lack of availability of a robust "gold standard" comparator for adherence in the vast majority of literature reviewed. The terms used were (asthma OR asthmatic) AND (adherence OR compliance OR concordance) AND (oral corticosteroids OR inhaled corticosteroids) AND (exhaled nitric oxide OR FeNO). We screened the entire reference list from each eligible study, and Google Scholar was used to find any relevant citing articles. The study selection was performed based on the following inclusion and exclusion criteria. 2. Studies published in full, between January 1975 and July 2020 3. Primary research involving direct corticosteroid monitoring or the level of exhaled nitric oxide as a marker of adherence, or at least was investigated with an assessment of the rate of adherence.

| Study identification and data extraction
The first author screened all the titles and abstracts to exclude nonrelevant articles. All the selected studies were screened in full text to assess the eligibility. The last author (SF) confirmed the eligibility of K E Y W O R D S adherence, asthma, corticosteroids, fractional exhaled nitric oxide studies and verified appropriate data extraction. Any disagreement between the authors in study selection and data extraction was resolved by discussion.

| Quality assessment
We used the National Heart, Lung, and Blood Institute of the National Institutes of Health (NIH) quality assessment tools for observational, cohort and cross-sectional studies, and controlled intervention studies. 20 The first (FA) and the second (AP) reviewers evaluated the included studies individually.

| Synthesis of results
Due to the different methodologies between the included studies, a meta-analysis was not conducted.

| RE SULTS
We identified 4603 studies through all databases. After removing duplicates, the titles and abstracts of 2850 articles were screened.
The full texts of 40 potential articles were retrieved for further evaluation; 21 fulfilled the inclusion criteria. Reference list checking and screening Google Scholar added five studies, resulting in 26 studies included in total ( Figure 1). Excluded studies are detailed in Table S1.
The relevant characteristics and findings of the selected articles are reported in Tables 1-5. The studies were conducted in 10 countries: 15 in the United Kingdom and Ireland, three in the Netherlands, two in the USA, and one each in Germany, Denmark, Spain, Finland, Greece and China. No relevant studies were found before 1998. The sample size of the included studies ranged from 17 to 529 participants.
Thirteen studies included only children, 12 adults and only one both.

| Quality assessment
Cohort and cross-sectional designs were most common (n = 22); the remaining four studies were randomized controlled trials (RCTs).

| Methods used for corticosteroid detection
Oral prednisolone was the only OCS used among the included studies, which typically reported blood prednisolone level in addition to cortisol, 9,16,17,27,28,30,37 applying various cut-offs to each as shown in Tables 1 and 2. One study reported ICS levels over 8 h after witnessed inhalation. 18 The analyses of prednisolone and

| Adherence rates by direct corticosteroid detection
Four studies investigated a daily dose of 40 mg prednisolone given as an "interventional" drug (Table 1). Overall poor adherence rates were between 8% and 25%. 16,17,27,30 Four further studies reported "real-world adherence" in patients who were prescribed regular daily prednisolone at enrolment. 9 We only considered those who have participated in the adherence measurements (not the whole participants of the study).

| FeNO suppression test as a marker of adherence
Two studies reported an application for FeNO testing in assessing adherence to ICS in patients with severe or uncontrolled asthma. 14

| D ISCUSS I ON
We have reviewed studies using objective biological methods for adherence monitoring in asthma, specifically through drug concentration assays in blood or measurement of exhaled nitric oxide. Our key findings are as follows: (1)  Most of these studies reporting direct serum monitoring were of poor or fair quality, with small sample sizes and cross-sectional designs. Prednisolone administration was almost always unsupervised, and so it is unknown whether lack of detection represents non-adherence or failure of the test. As expected, low adherence rates were more common in real-world study designs (20%-50%) when compared to interventional studies (8%-25%). Whilst most guidelines acknowledge that adherence should be assessed, they do not suggest a cut-off for "good" adherence. A cut-off of 80% has been suggested for example by McNichol et al. 14 and adopted widely (in the UK at least) as the standard required before escalation to biologics, with little supporting evidence. It may be that some patients "self-titrate" their treatment to maintain good control, and apparent "low adherence" may not be very important in such cases. However when a patient with asthma presents with an exacerbation or poor control, the investigation must include an appraisal of whether the current treatment plan is working, and for that of course, we need to know whether it is being followed.
Most studies reporting a single measurement of FeNO, 22,23,31,34 or repeated measures, 12,29,33,35 found higher levels in poorly adherent subjects. There are not enough data available to be able to propose a single cut-off, or a clinically meaningful change (fall) over time, for clinical use. This may be in part because important confounding factors such as age, height, gender and atopy were not taken into consideration. 48 In addition, the "gold standard" adherence in these studies often had significant limitations. A number of studies measured the level of adherence by diaries or self-report questionnaires. 22,23,29,31,32 The accuracy and reliability of such methods are low, 49 with patients typically over-estimating their adherence. 5,50,51 The FeNO suppression test may be able to identify patients with previous poor adherence, 14 and may be useful in a clinical setting at least in patients with elevated FeNO levels (>45 ppb), 14,25 where adequate resources are available. Further, non-suppressors may be identified with relative corticosteroid insensitivity who may require alternative therapeutic strategies.
Given the costs and time associated with adherence monitoring in asthma, it may be worth considering applying these methods only in selected patients where financial and clinical value will be most likely.
The most obvious area would be in those with apparent severe disease who are being considered for biologics, where better ICS adherence may improve disease control without the need for such high-cost therapies. In such patients, a relatively comprehensive assessment including pick-up checking, direct monitoring (blood ICS/OCS levels), smart inhalers (when available) and FeNO suppression (where relevant) may be justified. Conversely, it may also prove useful at start of asthma therapy, where guidelines suggest a trial of ICS therapy as part of the diagnostic workup 2 ; interpreting response without any knowledge of adherence is impossible and may lead to underdiagnoses or inappropriate therapy escalation (where there appears to be no response, but in fact, enough ICS has not been taken). It is here that the possibilities raised by the emergence of smart inhalers offer promise both in helping the clinician assess adherence patterns and potentially in helping the patient to establish adherent behaviours.
Several limitations of this review need to be considered. The included studies were selected based on the method of adherence from either direct drug measurement or measuring the FeNO level, and these methods cannot be compared with each other. Most studies were of low quality with high heterogeneity across the studies, including patient characteristics (in particular, asthma severity and age) and cut-off values of both adherence methods. Methods not included in the review, but of potential use, include the detection of corticosteroid metabolites in the urine, as reported in Ref. 52 In this study, fluticasone propionate-17beta-carboxylic acid was found in the urine of all 30 included subjects 16-24 h after witnessed inhalation of fluticasone propionate. A further single case report also found that fluticasone propionate and beclomethasone dipropionate were detectable in an induced sputum sample. 53 Further improvements to methods that assess the hypothalamic-pituitary-adrenal (HPA) axis may also prove useful; Smy et al. 54 found that hair cortisol during ICS treatment was reduced, and proposed that this as a useful surrogate marker of adherence. Finally, the robustness of this review would have been improved by independent searching, study selection and data extraction involving at least two researchers; here, one author performed these duties, and the results of each were subsequently verified by a second author.
In conclusion, we have reviewed the use of serum exogenous corticosteroid and cortisol detection, and of FeNO, as biological measures of adherence in asthma. Further work is required to adequately define the influence of drug dose-and timing-dependent factors on prednisolone and cortisol levels, in order to support their use in clinical and research practice. Although FeNO is usually lower in adherent patients, there are no data available to allow a single cut-off to be proposed. FeNO suppression testing merits further investigation as a marker of both adherence and steroid insensitivity.

ACK N OWLED G EM ENTS
We would like to thank the medical library team at Manchester

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.

AUTH O R CO NTR I B UTI O N
FA, SJF, BK and RN conceived and designed the analysis, and drafted the manuscript; FA and AP performed the searches and assessed data quality; all authors revised the draft manuscript and approved the final accepted version.

DATA AVA I L A B I L I T Y S TAT E M E N T
Data sharing is not applicable to this article as no new data were created or analysed in this study.