Remotely Monitored Therapy and Nitric Oxide Suppression Identifies Non-Adherence in Severe Asthma

Rationale Poor adherence is common in difficult-to-control asthma. Distinguishing patients with difficult-to-control asthma who respond to inhaled corticosteroids (ICS) from refractory asthma is an important clinical challenge. Objectives Suppression of fractional exhaled nitric oxide (FeNO) with directly observed ICS therapy over 7 days can identify non-adherence to ICS treatment in difficult-to-control asthma. We examined the feasibility and utility of FeNO suppression testing in routine clinical care within UK severe asthma centres using remote monitoring technologies.


Rationale
Poor adherence is common in difficult-to-control asthma. Distinguishing patients with difficult-tocontrol asthma who respond to inhaled corticosteroids (ICS) from refractory asthma is an important clinical challenge.

Objectives
Suppression of fractional exhaled nitric oxide (FeNO) with directly observed ICS therapy over 7 days can identify non-adherence to ICS treatment in difficult-to-control asthma. We examined the feasibility and utility of FeNO suppression testing in routine clinical care within UK severe asthma centres using remote monitoring technologies.

Methods
A web-based interface with integrated remote monitoring technology was developed to deliver FeNO suppression testing. We examined the utility of FeNO suppression testing to demonstrate ICS responsiveness and clinical benefit on electronically-monitored treatment with standard high dose ICS and long-acting β2-agonist (LABA) treatment.

Introduction
Non-adherence with inhaled corticosteroids (ICS) in poorly controlled severe asthma is common and associated with worse clinical outcomes 1,2,3 . Assessment of adherence to inhaled treatments is challenging: physician estimate and patient self-report consistently overestimate adherence when compared to objective measures, and other surrogate measures such as prescription collection records have limitations 4,5 . The advent of novel biologic therapies targeting type-2 cytokines in severe asthma makes identification of sub-optimal adherence to ICS important so that the issue can be addressed prior to committing a patient to a long-term parenteral therapy. Inhaler monitoring technologies remain the gold standard in clinical trials but clinical services have been slow to embrace these as part of routine care. They are perceived as potentially expensive and cumbersome, and there may be concern from healthcare professionals to challenge non-adherent behaviour or engage subjects to provide long term meaningful change in adherence behaviour.
Suppression of fractional exhaled nitric oxide (FeNO) is an easily measured predictor of ICS response 6,7,8 . Directly observed ICS treatment over seven days in subjects with a high FeNO (FeNO ≥45ppb) can identify subjects with difficult-to-control severe asthma who are responsive to ICS and non-adherent with their maintenance ICS treatment (FeNO suppression test [FeNOSuppT] -further detail see online supplement) 9 . A value of ≥45ppb was chosen because in subjects with severe asthma, it is associated with frequent exacerbations 10 and it identifies subjects who respond to treatment with ICS 6,11 . Thus, in poorly controlled severe asthma and FeNO ≥45ppb, the key clinical issue is whether subjects are taking their ICS treatment effectively or whether they need treatment escalation.
The aim of this work was to assess the utility of home-based delivery of the FeNOsuppT using a remote monitoring technology that assess both time of use and inhaler technique thereby enabling "directly observed therapy" to identify subjects who would achieve good asthma control with better inhaler technique or adherence to inhaled treatment. Some of the results of this study has been previously reported in the form of an abstract 12 .

Methods
This was a prospective evaluation of the FeNOsuppT in subjects attending UK severe asthma centres involved in the UK Refractory Asthma Stratification Programme (RASP-UK) 13 14 . This acoustic monitoring technology provides a time stamped sound file which can be analysed using an automated validated algorithm providing information on the timing and the technique of inhaler usage (figure e2, online supplement) 14 .
FeNO was measured using the Niox VERO™ and all clinical services were provided with the inhaler monitoring and FeNO technology with appropriate training for clinical staff in its use.
Subjects considered for FeNOsuppT were referrals to a severe asthma clinic with poor asthma control despite prescription of high dose ICS in combination with long-acting β2-agonist (ICS/LABA) and an elevated FeNO (defined as FeNO≥45ppbhigh FeNO group). Subjects were asked to measure their FeNO daily and to take high dose ICS for seven days (1000 µg fluticasone per day) via the Diskus™ with an INCA™ device, in addition to their usual ICA/LABA. Printed instruction sheets for the inhaler and the Niox VERO were given to each patient.
On return to the clinic, results from both the INCA™ device and the Niox VERO™ were uploaded to a server using the Vitalograph IC Data Compression Utility application and analysed using the embedded sound analysis algorithm (Figure e3, online supplement). A FeNOsuppT test was defined as previously 9 but in effect, a 42% fall in FENO between the Day 0/Day 1 and the Day 4/Day 5 mean values equates to a positive suppression test. If a patient could use the Diskus device efficiently based on the FeNOSuppT monitoring profile, they were offered the opportunity to replace their current ICS/LABA with salmeterol 50µg/fluticasone 500µg Diskus one inhalation twice per day with an INCA™ device, and were informed that inhaler use was being monitored for a one month period (Figure e4, online supplement), After one month of monitoring, the relationship between the FeNOsuppT and biomarker (FeNO and peripheral blood eosinophil count) and clinical outcomes (Asthma Control Questionnaire (ACQ-5) and spirometry) in subjects with good adherence was examined. Good adherence was defined as ≥70% adherence using the INCA™ device as previously in a clinical trial setting, mean adherence was 73% using biofeedback and thus likely to be best achievable in a real-world setting 14 .
In parallel, a sub-group of subjects with poor asthma control despite high dose ICS/LABA treatment and FeNO<45ppb at one clinical centre (Belfast) (low FeNO group), were offered INCA™ monitoring. They were instructed in the use of Accuhaler device, and if they could use it efficiently (based on use of a training device at clinic), they were provided with salmeterol (50µg/fluticasone 500µg Diskus one inhalation twice per day) with an INCA™ device, to be taken in the morning and in the evening for one month.

Statistical analysis
Data were anonymised and collected in a systematic manner to allow reporting of the utility of the testing in routine clinical care. Normally distributed variables data were presented using the mean  standard deviation. Mean differences were calculated with 95% confidence intervals, and independent and paired t-tests were used to formally compare between groups. Non-normally distributed variables were presented using the median and the interquartile range. Median differences were calculated with bootstrapped 95% confidence intervals, and the Mann-Whitney U test or Wilcoxon signed-rank test were used to formally compare between groups. Chi-squared tests were used to test for differences across groups for categorical variables. Correlation analyses were conducted using the Spearman's rank test. All statistical analyses were performed using STATA version 14.

Results
Patient disposition and demographic details for high and low FeNO groups are shown in Figure 1 and Table e1 in the online supplement.

FeNO Suppression test
Of 250 subjects with FeNO≥45ppb who performed a FeNOsuppT, 49 of 250 (20%) subjects were unable to perform the test -16 (6%) did not measure their FeNO daily and 33 (13%) had critical inhalation errors or missed multiple doses (<70% of ICS during FeNOSuppT) of the additional inhaler over a single week. This was despite awareness that inhaler usage was being monitored and despite detailed verbal and written instructions, and a proven initial ability to measure their FeNO and use the Diskus efficiently in the clinic. FeNO at the start of the test, but were younger (p<0·001), more likely to be female (p=0.007), had higher rates of atopic eczema (p<0·001), higher IgE levels (p=0.012) and were less likely to be on maintenance prednisolone (p=0·025) ( Table 1).
There was no difference in baseline FeNO in subjects with a positive suppression test on maintenance prednisolone (median 84ppb, IQR 65, 111) compared to subjects not taking prednisolone (median 92ppb, IQR 63-127) whereas blood eosinophils were significantly lower in subjects on prednisolone (median 0.40 x10 9 /L, IQR 0.19, 0.61 versus 0.56 x10 9 /L, IQR 0.30-0.84, p<0.01). In subjects prescribed maintenance prednisolone, 61 (58%) had a positive suppression test and 45 (42%) had a negative test and there was an identical pattern of FeNO suppression in subjects taking prednisolone compared with subjects not taking prednisolone with no differences in baseline, Day 4 or Day 7 FeNO levels (figure e6 online supplement). This was also was the case in those subjects on maintenance prednisolone with contemporaneous positive prednisolone / cortisol levels on Day 0 (n=59, data not shown).
Blood eosinophil counts (figure 3a and 3b) fell significantly from Day 0 to Day 7 in those subjects who had a positive FeNOsuppT [median difference = -0.14 (95% CI -0.03, -0.23), p<0.001] but not in subjects with a negative suppression test [median difference = 0.02 (95% CI -0.12, 0.14),  assessed by severe asthma services with difficult-to-control asthma (n=290). Subjects with FeNO≥45ppb (multicentre high FeNO group, n=250) had FeNO suppression testing and if they liked the Diskus and could use the device proficiently proceeded to one month monitored treatment with high dose ICS/LABA (salmeterol 50µg/fluticasone 500µg Diskus one inhalation twice per day). Subjects with FeNO<45ppb (single centre low FeNO group, n=40) had identical one month monitored treatment.

Post one month monitoring after FeNO suppression test
Of the subjects who successfully performed suppression testing, 130 subjects agreed to proceed to further monitoring for one month to assist with use of their inhaled treatment 89 (68%) with a positive test and 41 (58%) with a negative test ( There was a strong relationship between FeNO at Day 7 of the FeNOsuppT and FeNO level after one month monitored treatment with good adherence which was also the case for subjects on maintenance prednisolone ( figure 4). There was also a significant reduction in blood eosinophil  Figure 6 (FeNO, r=0.69; p<0·001, blood eosinophils r=0.60; p<0·001) with post one month monitoring FeNO generally lower than day-7 FeNO value.

Predictive value of FeNO Suppression test
We also examined the predictive value of a positive 7-day suppression test for FeNO level in subjects with good adherence after one month monitoring, using Receiver Operating Characteristic (ROC) analysis (Figure e7, online supplement). The area under the curve was 0·81 (95% CI, 0·72 -0·91) and the optimal cut-point for specificity and sensitivity was FeNO 35ppb (in severe asthma, FeNO<35ppb has been shown to be associated with significantly less airway reactivity, airflow limitation and hyperinflation and significantly reduced emergency room and intensive care unit admissions 15 ).

FeNO low group
In the low FeNO group (<45ppb), who proceeded to monitoring for one month (n=40,

Lung function and asthma symptom scores
Lung function data and ACQ-5 scores at baseline and post one month monitoring in all subjects with good adherence are shown in Table 3

Discussion
This study demonstrates in subjects with difficult-to-control severe asthma and FeNO ≥45ppb, 65% had a positive FeNOsuppT and with effective adherence to LABA/ICS during a one month monitored period, there were significant improvements in both symptoms and lung function and FeNO was maintained at target levels associated with reduced exacerbation, emergency room attendances and ICU admissions 15,16 . In contrast, among those with a negative suppression test, there was a lesser reduction in FeNO despite taking high dose ICS/LABA efficiently, with no improvement in lung function and less improvement in symptoms.
The original description of the FeNOSuppT previously used physician directly-observed treatment (DOT) over 7 days and defined a positive FeNOsuppT for non-adherence on poor prescription filling. The remote monitoring acoustic technology in this study effectively delivers DOT by timestamping inhaler activation and analysing inhalation technique which also identifies "non-intentional" non-adherence, where critical inhaler errors prevent effective treatment which is known to be a common problem 17 . Given these patients were prescribed high dose ICS/LABA at the time of FeNOsuppT, we believe the suppression of FeNO with effectively administered ICS/LABA treatment reflects prior inefficient ICS treatment due to both intentional and non-intentional nonadherence.
In subjects with a positive FeNOsuppT, there was also a significant fall in blood eosinophil counts consistent with both biomarkers being ICS responsive when adherent with monitored treatment.
Composite biomarker profiling using FeNO and blood eosinophil count allows better prognostic risk stratification with highest risk seen when both Type-2 biomarkers are high and vice versa 18,19 , suggesting that for subjects with a positive FeNOsuppT and resultant biomarker low profile with good adherence, there should be a parallel risk reduction for exacerbations. The longer term clinical and adherence outcomes for subjects characterised using FeNOsuppT in a severe asthma population are currently being studied (ClinicalTrials.Gov NCT02307669), but it is worth commenting that a small number of subjects, despite a good response to monitored treatment in terms of FeNO, ACQ-5 and lung function improvement, have a persistent elevated peripheral blood eosinophil count. This suggests that the biological driver of blood eosinophils in these subjects may not be as ICS responsive as FeNO, and we are also currently exploring the clinical significance of this "dissociated" biomarker profile in terms of exacerbation risk and clinical outcome (ClinicalTrials.Gov NCT02717689). However, our data suggest that the biomarker profile (FeNO and blood eosinophil count) after 1 week of FeNOsuppT is closely related to the profile when taking optimised ICS/LABA treatment, thus facilitating identification of subjects where adherence intervention to optimise inhaled treatment may be of most value prior to considering treatment escalation but also potentially identifying those subjects who are likely remain biomarker high despite optimised ICS/LABA treatment and suitable for novel type 2 biologic therapies. Maximal FeNO suppression has also been shown after 7 days high dose ICS treatment in mild asthma 7,20,21 and our data suggests that similar maximal suppression is seen in subjects with more severe asthma given the strong correlation between day 7 FeNO post FeNOSuppT and post monitoring FeNO. In those subjects despite having with a positive test and FeNO >35ppb after monitoring, significant falls in both FeNO was seen compared to baseline with improvements in lung function and symptom scores consistent with prior ICS non-adherence and suggesting a period of optimised inhaled treatment with clinical monitoring prior to treatment escalation is appropriate.
Poor-adherence is common in subjects with severe asthma 1.2,3 . The availability of novel biologic therapies targeting type-2 cytokines makes identification of sub-optimal adherence to ICS important so that these issues can be explored prior to treatment escalation. An attractive proposition is to move away from using imperfect surrogate measures of adherence (e.g. patient self-report, physician impression, prescription records) to using an ICS responsive biomarker (FeNO) as part of a clinical phenotyping strategy and precision medicine delivery. The FeNO-low group (FeNO<45ppb) demonstrate a small reduction in FeNO with monitored optimised treatment with ICS/LABA with a FeNO in a similar range to those subjects with a positive suppression test and optimised treatment. There was no change in peripheral blood eosinophil count, ACQ-5 or lung function, which may reflect the fact that this low FeNO group are more adherent to background ICS treatment and the potential for further clinical improvement with optimised ICS/LABA in this FeNO-low group is less.
A strength of this data is that it occurred in a "real world" setting, by-passing the challenging aspect of engaging a non-adherent difficult-to-control asthma population in a clinical trial 22 . As well as providing insight into ICS therapeutic response in a short period of time, this approach identifies subjects who can use a particular inhaler efficiently and wish to engage with a monitoring strategy to assist them with their inhaler use. However, it is worth emphasising that 70 of 169 subjects [41%] failed to take ≥70% or complete 1 month monitoring of inhaled treatment despite being proficient in inhaler use and keen for support, suggesting there are barriers to even short-term use of technological support to assist with inhaled treatment adherence. If these barriers could be identified and addressed, there could be substantial gains with optimised use of high dose ICS/LABA treatment in this poorly controlled, high-risk asthma population, however defining the best intervention to change non-adherent behaviour is outwith the scope of this study. Only 16 patients were unable or forgot to perform daily FeNO during the 7-day home assessment suggesting that this is not a major barrier to home delivery of domiciliary FeNOSuppT, however inadequate use of the monitored ICS was more common.
Formal cost-effectiveness analysis of 'biomarker-based' assessments with adherence monitoring is also required, however this seems likely by identifying subjects who may achieve adequate asthma control without the need for treatment escalation to expensive biologic agents. Subjects with a positive FeNOSuppT were younger with more eczema and higher IgE suggesting greater ICS responsiveness in this group though hospital admissions were equally prevalent in the high FeNO group with positive and negative suppression tests. In the National Heart, Lung, and Blood Institute Severe Asthma Research Program, a high FeNO (>35ppb) was associated with greater emergency room visits and hospital and ICU admissions and again in a younger, more atopic patient cohort 15 .
If adherence could be satisfactorily addressed in this ICS responsive FeNO high difficult-to-control asthma population, there could potentially be a substantial impact on hospital admission rates.
In summary, we have demonstrated that and FeNOsuppT can be delivered using remote monitoring technology in routine clinical care in specialist severe asthma services in the UK. Short-term profiling of the FeNO response to ICS exposure in subjects with difficult-to-control severe asthma is a helpful part of clinical phenotyping to identify those who are likely to respond better to high dose ICS/LABA therapy when used regularly and those who, despite good adherence with inhaled treatment, are likely to require additional treatment. In addition, aligning monitoring of ICS treatment with FeNO response may be of value in identifying ICS responsive subjects prior to recruitment to clinical trials investigating interventions which are "add-on" treatments to standard care (ICS/LABA) and also to studies investigating "true" ICS resistant disease. Future studies are required to identify the optimal intervention to maintain ongoing adherence in a difficult-to-control severe asthma population, but understanding the potential therapeutic benefit in an individual patient is a useful part of routine care.

Acoustic Monitoring Technology
The INCA™ device is attached to the Accuhaler and a microphone is activated when the device is opened and stops when the inhaler device is closed (Figure e1). If the inhaler device is left open, the recoding automatically shuts off after 90 seconds.
The INCA™ Analysis software processes the sound files generated by the INCA™ device to assess if the inhaler has been taken properly (Figure e2). Most of the inhalations during the service evaluation were deemed satisfactory by the software and any "technique errors" identified by the software, were subsequently manually over-read by the clinical team.
Feedback from the 7 day Flixotide Accuhaler technique and usage (Figure e3) during the FeNO suppression testing was also examined to ensure that patients could use the Accuhaler device efficiently. If this was the case and patients wished to continue to use the technology to assist with their inhaled treatment, they were then provided with a salmeterol 50µg/fluticasone 500µg Accuhaler with INCA™ device to be taken once in the morning and once in the evening (standard high dose ICS/LABA therapy) and again instructed in inhaler use. On return to the clinic, the data was uploaded and an easily interpretable readout was produced for the clinician to provide further feedback to the patient (figure e4).