Short- and medium-term effect of inhaled corticosteroids on exhaled breath biomarkers in severe asthma

Fahad H Alahmadi; Max Wilkinson; Brian Keevil; Rob Niven; Stephen J Fowler

doi:10.1088/1752-7163/ac7a57

1. Introduction

Asthma is the most common chronic inflammatory lung disease, characterised by paroxysmal airway inflammation and smooth muscle hyper-responsiveness resulting in variable airway obstruction. Regular inhaled corticosteroids (ICS) and as required β₂-agonists are the primary therapy in managing and controlling asthma, although responsiveness to treatment varies due to the heterogeneity of the disease.

Since several different inflammatory phenotypes have been identified in asthma, pathological and clinical features have been investigated regarding ICS response. It has been shown that after the ICS administration, patients have reduced levels of sputum eosinophils [1, 2], which then increase once treatment is withdrawn [3] or stepped down [4]. Recently, fractional exhaled nitric oxide (FeNO) has been used to monitor ICS response in asthma at all levels of severity [5–7]. The level of FeNO in mild asthma drops when corticosteroid medication is administered and rises when steroids are discontinued [8]. In contrast, FeNO levels remain elevated in some patients with severe asthma despite being prescribed high dose ICS [9], indicating non-adherence or, less commonly, relative steroid insensitivity. There is interest in other non-invasive biomarkers to supplement FeNO in predicting steroid treatment responsiveness and aid in phenotyping asthma.

One of the features of airway inflammation is increased bronchial blood flow, caused by mediators such as histamine, bradykinin, and nitric oxide. Asthmatics have a higher exhaled breath temperature (EBT) when compared to healthy controls [10]. EBT has been evaluated to assess steroid response in asthma, although the results are inconclusive, with a non-significant reduction in breath temperature 30 min post-ICS administration [11].

More recently, the measurement of particles in exhaled air (PExA) has been developed as a non-invasive tool by Almstrand et al [12] sampling particles which arise from the peripheral airways. An exhaled aerosol is collected that comprises liquid particles of respiratory tract lining fluid formed in the small airways when performing a breathing manoeuvre that induces airway closure and re-opening [13–15]. Asthmatics exposed to birch pollen to induce inflammation have been found to have reduced exhaled particle count after exposure [16], postulated to be due to the adverse effect of inflammation on the repeated small airway opening and closure that is thought to occur during the manoeuvre.

Exhaled volatile organic compounds (VOCs) that originate from both endogenous and exogenous sources may provide an alternative method for monitoring airway inflammation [17] and assessing asthma control [18]. Schee et al studied the response to oral prednisone in patients with mild/moderate asthma and found that VOC patterns differed between steroid-responsive and steroid-unresponsive patients [19]. Analysis of exhaled VOCs could therefore identify inhaled steroid response and be used to assess ICS adherence.

In this study we aimed to investigate whether exhaled breath tests can be used as surrogate markers of ICS response in severe asthma patients by using a range of methodologies: FeNO, EBT, PExA and VOCs.

2. Methods

2.1. Patient selection and setting

Patients attending the severe asthma clinic at Wythenshawe Hospital, Manchester, UK, with an elevated FeNO (⩾45 ppb) were recruited. Participants were 18 years or older, non-smokers with uncontrolled asthma symptoms and treated with high dose of ICS and another controller.

2.2. Study design

A FeNO suppression test was performed, whereby participants were prescribed high dose ICS (Flixotide 500 Accuhaler, two puffs daily for 1 wk) with the inhaler dosing monitored using an INCA device (Vitalograph, UK), which recorded date and time of actuation and inhaler technique (sufficient flow and time of inhalation). This allowed the clinical team to assess whether the previously high FeNO was due to poor ICS adherence (in which case the FeNO would be suppressed after a few days of monitored dosing), or relative corticosteroid insensitivity (at least in terms of FeNO, where FeNO would not be suppressed despite regular inhalation of high dose ICS).

All participants gave written informed consent. The study received ethical approval from the South-East Exeter research ethics committee (REC reference: 18/SW/0058). The study was conducted during two visits over 1 wk. At the baseline visit, full medical history was obtained, asthma control measured used the asthma control questionnaire (ACQ) [20], and spirometry performed according to European Respiratory Society guidelines [21]. Then, each patient inhaled two puffs of Flixotide 500 Accuhaler, immediately followed by measurements of (in sequence) VOCs, EBT, FeNO, and (at baseline, 120 min and day 7) PExA. These exhaled measurements were repeated after 30, 60 and 120 min, and again at day 7. VOC measurements were performed first, to minimise any potential contamination from the instruments used for the other tests, and PExA done last to avoid any effect of repeated forced exhalation manoeuvres on the results [22]. Spirometry and ACQ were also repeated at day 7. The full study design and sequence of procedures are presented in figure 1.

**Figure 1.** Time and order of study procedures.
Download figure:
Standard image High-resolution image

2.3. Measurements

2.3.1. Fractional exhaled nitric oxide (FeNO)

FeNO tests were performed on four occasions during the first visit, and at visit 2, using the portable nitric oxide analyser Niox Mino (Aerocrine AB, Solana, Sweden) at an exhalation flow rate of 0.05 l s⁻¹. FeNO measurements were performed according to American Thoracic Society recommendations [5].

2.3.2. Exhaled breath temperature (EBT)

EBT was measured by the X-Halo device (Delmedica) which has been validated by Poppa et al [23]. Before EBT measurements were performed, body and room temperature were recorded. During sampling the device was connected through a cable to a mobile phone loaded with the relevant application and connected to the internet. The manufacturer of X-Halo recommends that the device should not be used immediately after consuming any food, so patients were asked to avoid eating for 1 h before the baseline test [24]. Sampling was conducted according to the following procedure: patients were instructed to inhale freely through their nose and exhale through the mouthpiece of the device whilst maintaining a normal tidal breathing rhythm. During sampling they were instructed to keep their mouth on the sampling nozzle until the device indicated the test was complete. The period of the manoeuvre was approximately 1 min.

2.3.3. Exhaled volatile organic compounds (VOCs)

Exhaled VOC samples were collected from each subject by using the Respiration Collector for In Vitro Analysis (ReCIVA, Owlstone, Cambridge, UK). The device was connected to a Clean Air Supply Pump (CASPER) for ReCIVA which provided a continuous flow of 40 l min⁻¹ of VOC free air to the ReCIVA. During sampling, the ReCIVA directed the airflow from the exhaled breath onto four sorbent tubes packed with Carbograph 1TD/Carbograph 5TD (Markes International, Llantrisant, UK). The ReCIVA allowed the collection of a specific breath fraction by constant monitoring of the pressure inside the mask. This ensured only the end tidal portion of breath was collected, minimising contamination from the mouth and airway deadspace. Prior to use thermal desorption (TD) tubes were conditioned using a TC20 (Markes International, Llantrisant, UK) for 60 min at 320 °C. The TD tubes were sealed before and after sampling to prevent contamination with exogenous compounds.

Before sampling, background air samples were collected to assess any VOC contamination from the breath sampling setup (e.g. CASPER air supply or the face mask). This reference sample was collected by strapping the ReCIVA to a glass head and setting the pumps to 'always on' allowing the collection of 500 ml of gas at a flow rate 200 ml min⁻¹.

During sampling patients were advised to perform regular tidal breathing to allow the collection of 500 ml of gas at a flow rate 200 ml min⁻¹ which took between 6 and 10 min depending on breathing rate. Once sampling was completed the sorbent tubes were removed from the ReCIVA to be recorded on the CRF and stored immediately in the refrigerator at 4 °C.

Breath sample analysis: Samples were analysed at the University of Manchester, Manchester, United Kingdom. All tubes were dry purged for 7 min in a flow of clean N₂ at 50 ml min⁻¹ using a TC-20 (Markes International). Samples were loaded with an internal standard (1 ppmv, 4-Bromofluorobenzene in nitrogen, Thames Restek, UK) and desorbed using a TD-100 (Markes International) onto a 7890B gas chromatograph (Agilent, SantaClara, CA, USA) connected to a 7010 (Agilent) triple quadrupole mass spectrometer. The parameters for desorption and analysis have been described in full elsewhere [25]. Files were processed using Agilent Masshunter to deconvolve and align peaks within the chromatograms.

2.3.4. Particles in exhaled air (PExA)

The exhaled particle (PEx) mass was collected using a PExA 2.0 device (PExA, Gothenburg, Sweden). All the participants were trained to do the PExA breath manoeuvre. Firstly, participants wore nose clips and breathed normally in and out via a mouthpiece connected to a two-way valve in the PExA instrument. After that, patients emptied their lungs through a deep exhalation until the residual volume and held their breath for five seconds. Patients then inspired rapidly to their total lung capacity and then breathed out steadily (maximum exhalation flow of 2000 l s⁻¹). Patients were asked to breathe normally for 2 min until the next breath sample. During this time they were provided filtered air to prevent any contamination from the ambient air. Patients were asked to repeat this manoeuvre until a target mass of PEx had been collected (50 ng). Normally 10–15 manoeuvres were performed by each participant to collect at least 50–100 ng. However, there were a few patients who were unable to provide sufficient mass after 15 repeated manoeuvres; these individuals were not excluded from the PExA data.

Surfactant protein A (SPA) and albumin detection in PEx: All samples were kept in a cryotubes and stored at −80 °C until the analysis. After the last patient was sampled, the samples were sent from North West Lung Centre Research laboratory at Wythenshawe hospital to Sahlgrenska Academy, University of Gothenburg, Sweden for lipids and proteins analysis.

An extraction buffer of 10 mM of phosphate-buffered saline containing 1% bovine serum albumin w/v and 0.05% Tween® 20 (Thermo Scientific, Rockford, IL, USA) was prepared. Particle extraction was performed using 160 μl of buffer for whole PEx filters. The extract was incubated at 37 °C for an hour while being shaken at 400 rpm (Eppendorf Thermomixer comfort, Eppendorf AG, Hamburg, Germany).

Aliquots of 40 µl for SPA and albumin analysis were taken with the remainder retained as a backup sample. Prior to enzyme-linked immunosorbent assay (ELISA) analysis the extracted PEx samples were stored at −20 °C for one and two days for the SPA and albumin aliquots respectively. For both the SPA ELISA (BioVendor, Brno, Czech Republic) and human albumin ELISA (Immunology Consultants Laboratory, Inc. Portland, USA) an 80 μl assay dilution buffer was added to the samples prior to analysis. In order to unify the sample matrix, a 1:2 ratio was prepared between extraction buffer and assay dilution buffer in all PEx samples, controls and standard samples [26].

After addition of the assay dilution buffer SPA samples were incubated at 37 °C for 2 h whilst being shaken at 300 rpm, and albumin samples at room temperature for 1 h whilst being shaken at 300 rpm. A reaction time of 9 min was used for both assays. The coefficient of variation was 0.5 ng ml⁻¹ for the SPA ELISA and 0.9 for the albumin ELISA. The limit of quantification for the SPA and albumin tests were 0.1 and 0.8 (ng l⁻¹) respectively.

2.4. Statistical analysis

This was an exploratory study assessing endpoints from several different methods, and as such no power calculation was performed. Statistical analysis was performed using SPSS version 23 (Chicago, IL, USA) and Graphpad Prism version 8 (San Diego, CA, USA). Comparison between repeated measurements at day 1 was done by Friedman test. Comparison of day 1 and day 7 samples was performed using a paired Wilcoxon test. Day 7 samples were collected between 09:00 and 10:00 in the morning and therefore fall within 1–2 h of a participants last dose of ICS. To ensure appropriate baseline (day 1) values were used the mean of all measurements taken in the 2 h post-ICS inhalation was taken.

Correlations were calculated using Spearman' correlation coefficient (r). Partial correlation coefficients were calculated between PEx and FeNO or FEV₁. In VOC analysis, data were normalised to the internal standard and log transformed. Despite this transformation many VOCs were still not normally distributed, and hence within- and between-day comparisons here made using the Wilcoxon test. A false discovery rate (FDR) of 0.1 was applied using the Benjamini–Hochberg method due to the large number of potential markers measured.

3. Results

3.1. Demographics

Seventeen patients with severe asthma and treated with at least 1000 mcg per day of beclometasone dipropionate (BDP) equivalent were recruited. Two patients were unable to maintain the steady and prolonged expiration required for FeNO, and three others could not tolerate the repeated forced exhalations for PExA; further data were not collected from these, and so demographics of the remaining 12 who continued in the study are summarised in table 1. There was no significant change in asthma control (measured by the ACQ) or FEV₁ over the week (p = 0.101 and 0.635 respectively).

Table 1. Participant characteristics, data shown as median (interquartile range (IQR)) except where indicated. ACQ: asthma control questionnaire (controlled asthma = score ⩽ 0.75 [27]); BMI: body mass index; FEV₁: forced expiratory rate in 1 s; FVC: forced vital capacity. NB all participants had severe asthma and as such were prescribed high dose ICS and at least one additional controller medication.

	Participants n = 12
Age, yrs	45 (36–59)
Female %	25
BMI, kg m⁻²	27 (23–29)
FEV₁, % predicted	70 (46–88)
FVC, % predicted	83 (58–98)
ACQ	2.4 (0.6–3.8)
Exacerbation rate, n yr⁻¹	1 (0–1)
Comorbidities, n	1 (0–1)
Smoking, n	Ex = 4 Non = 8

3.2. Fractional exhaled nitric oxide

The median (IQR) FeNO levels at baseline, 30 min, 60 min, and 120 min were 87 (59–124), 85 (68–137), 102 (66–143), and 97 (68–145) ppb respectively, with no significant variability in FeNO between these time points (p= 0.378). FeNO levels on the day 7 fell compared to day 1 [50 (31–71) ppb versus 87 (58–136) ppb, p= 0.009, figure 2]. Nine of the 12 participants had a significant (42% [28]) fall on FeNO over the week. FEV₁ at day 1 was negatively correlated with FeNO levels at both day 1 (r = −0.793 p = 0.004) and day 7 (r = −0.675 p = 0.032).

**Figure 2.** Box and whisker plot showing fractional exhaled nitric oxide (FeNO) levels at day 1 (mean of four readings over 2 h) and at day 7 for n = 12 patients.
Download figure:
Standard image High-resolution image

3.3. Exhaled breath temperature

One patient had EBT consistently below 31 °C at day 1—this was a clear outlier compared to the rest of the dataset, and compared to previous asthma data [29] and removed from subsequent analysis. For the remainder, there was significant change in EBT over 2 h post ICS administration (p = 0.017, figure 3). The median (IQR) temperatures at the baseline, 30 min, 60 min and 120 were 34.10 (33.50–34.53) °C, 33.74 (33.06–34.19) °C, 33.78 (33.06–34.12) °C and 33.86 (33.18–34.40) °C respectively, with significant variability in EBT values between these time points (Friedman-test, p = 0.017). All participants except two had reduced EBT at day 7; the median (IQR) temperatures were 34.03 (33.02–34.42) °C and 33.55 (32.33–33.88) °C for day 1 and day 7 respectively, and this was statistically significant (p = 0.017).

**Figure 3.** Box blot of EBT levels at different time point within 2 h (A) and mean over 2 h and at the day 7 (B). Note: We excluded one patient with EBT level below <31 °C.
Download figure:
Standard image High-resolution image

3.4. Particles in exhaled air

Ten participants each provided three PExA samples: two at day 1 and one at day 7. The median (IQR) of the PEx mass at baseline and after 2 h were 19.6 (7.2–64.2) ng and 25.4 (18.2–61.9) ng respectively (p= 0.575). Mean levels at day 1 were not significantly different compared to day 7 (table 2).

Table 2. PExA sampling parameters and protein concentration in all study visits. Data expressed as median (IQR). PEx: mass collected from exhaled particles; SPA concentration.

	Day 1: baseline	Day 1: 120 min	Day 7	p-value	Mean day 1 vs day 7 p value
Time (min)	10.5 (2.3)	10.7 (2.6)	11(3.7)	0.882	0.819
Mass PEx (ng)	19.6 (7.2–64.2)	25.4 (18.2–61.9)	47.5 (10.5–109.5)	0.607	0.575
Particles per exhalation	102 507 (560–164 007)	126 590 (53 811–243 112)	189 327 (30 727–305 091)	0.417	0.779
SPA (ng l⁻¹)	0.8 (0.3–2.0)	1.5 (0.7–1.9)	1.4 (0.5–2.5)	0.882	0.624
Albumin (ng l⁻¹)	0.6 (0.4–1.5)	1.4 (0.7–1.8)	1.5 (0.9–2.3)	0.417	0.263

There was no significant difference in the concentration of SPA and albumin between the mean of day 1 and day 7 (table 2). Further, there were no significant differences in the calculated weight percentage of SPA (p= 0.989) or albumin (p= 0.674) between day 1 and day 7.

The % predicted FEV₁ was correlated with the mean of generated PEx (r= 0.717, p= 0.030) and the mean particles per breath manoeuvre (r= 0.750, p= 0.020) for samples collected during the 2 h following ICS inhalation on day 1 (figure 4). However, this association was not observed at day 7. A possible (non-significant) negative relationship between the levels of FeNO in each visit and the collected PEx (ng) at baseline only (r = −0.612 p= 0.060, r = −0.635 p= 0.091 respectively) was also observed. The FEV₁% predicted and PEx were still highly correlated in the partial correlation analysis after adjusting for the effect of FeNO (r= 0.746 p= 0.089), while the correlation for PEx and FeNO was non-significant after adjusting for FEV₁ (r= −0.393 p= 0.384).

**Figure 4.** Top: Relationship between the mean level of collected PEx at day 1 and FEV₁ predicted (r= 0.717 p= 0.030). Bottom: Relationship between the mean number of particles per exhalation at day 1 and FEV₁ predicted (r = 0.750 p = 0.020).
Download figure:
Standard image High-resolution image

The mean measurements of PEx mass at the day 1 correlated with the levels of SPA (r = 0.806, p = 0.005) and albumin (r = 0.709, p = 0.002).

3.5. Volatile organic compounds

Data from one participant were not used as VOC samples were collected at day 7 only. A total of 139 VOCs were measured in the four samples from the baseline visit. Of these, 14 were significantly different between the time points with six having an FDR q< 0.1. Using the National Institute of Standards and Technology library three compounds were tentatively identified as ethyl benzene, o-xylene, and tetradecane (figure 5). Furfural and xylene were significantly differently expressed between the baseline and day seven visits (p < 0.05, figure 6) but neither of had FDR q < 0.1.

**Figure 5.** Box and whisker plots comparing measurements in day 1 for identified VOCs with FDR q < 0.1.
Download figure:
Standard image High-resolution image

**Figure 6.** Box and whisker plots comparing mean measurements of day 1 with day 7 (p < 0.05). Note these compounds have FDR above q > 0.1.
Download figure:
Standard image High-resolution image

4. Discussion

In this work we have investigated a broad range of non-invasive exhaled breath profiles as potential markers of ICS responsiveness. Four sample types were used, FeNO, EBT, PExA and VOCs. As expected a significant reduction in the FeNO levels was observed following seven days of ICS treatment, but the rapid impact of ICS on EBT was demonstrated here for the first time. Furthermore there was clear variability in the pattern of some VOCs following ICS use. In contrast, we do not observe any change in any of the PExA parameters. This study indicates that exhaled biomarkers (FeNO, EBT and VOCs) might be useful for evaluating short and/or medium-term ICS response, and larger studies that give treatment over a longer period to enable assessment of clinical effects (e.g. on symptoms and lung function) are now required

ICS are used to treat airway inflammation in asthma. It has previously been shown that FeNO is correlated with the degree of the inflammation and is an accurate predictor of ICS response. The results presented here are in line with previous literature that showed daily use of ICS suppresses FeNO [28, 30]. However, no difference in FeNO levels over the 2 h after ICS administration were seen; in fact the measurements were very similar and indicate high reproducibility of FeNO over this timeframe. This is believed to be the first work reporting on FeNO within a 2 h interval following ICS-dosing. The inverse correlation of FeNO values and % predicted FEV₁ shown here is also consistent other studies [31–33].

This work shows for the first time that there is a significant short-term reduction of EBT after ICS administration which was also observed after seven days of ICS use. It has been suggested the airway inflammation and remodelling increases bronchial vascularity and that this may lead to elevated heat exchange in the airway [10, 34]. ICS can cause acute reduction (within 30–90 min) in the bronchial blood flow [35], which may result in reduced airway heat. Thus, EBT may be clinically useful to assess acute responsiveness to ICS medication over a very short period. Moreover, since we have also observed a significant reduction in EBT levels after seven days of monitored ICS administration, it could merit further investigation as a possible marker of ICS adherence.

ICS medications did not show any significant effect on PEx protein biomarkers (SPA and albumin) over a short and medium time-period. These results are similar to Larsson et al who reported no difference between asthmatics using ICS and those not on ICS [16]. The acquired mass of proteins was lower than found in earlier studies [16, 36, 37], probably due to high correlation between the PEx mass and SPA and albumin levels, an association that has previously been reported [37]. Exhaled proteins are formed and transported via exhaled aerosol particles which are highly dependent on the participants breathing pattern [12, 13].

Despite the limitation of the low PEx mass collected, % predicted FEV₁ was strongly and positively correlated with PEx and particles per exhalation collected within the 2 h following inhalation of ICS; however, this was not observed after a week of continuous ICS use. Reduction in lung function is associated with air trapping and may reduce the closure and opening of the small airways required to produce PEx [38]. Further, the inflammatory processes that occur in asthmatic small airways and alveoli may change their mechanical characteristics and lead to airway closure and opening occurring more proximally, leading to fewer particles produced. On the other hand, a potential relationship was found between FeNO at both visits and PEx mass, although non-significant after adjusting for lung function, indicating that lung function is the main driver for PEx mass.

Changes in exhaled VOCs over both 2 h and 1 wk of ICS administration were demonstrated. Ethyl benzene, xylene, and tetradecane expression was shown to change within a short timeframe, and o-xylene and furfural levels differed between the two visits. Xylene and tetradecane have previously been shown to discriminate between asthma and healthy controls, and o-xylene compounds have been related to the level of asthma control [18, 39, 40]. This is the first assessment of these compounds in relation to ICS responsiveness suggesting their use as biomarkers but requiring further work to be done to confirm the change in these compounds after ICS use, including in dose-response studies.

The main strength of this study was that the use of ICS medication was recorded for each participant during the two visits. Secondly, most of the participants were proven to be steroid responsive, at least in terms of FeNO, with levels dropping at the second visit in 10 out of 12 cases. It should be noted though that FeNO suppression reflects only one aspect of corticosteroid sensitivity, in relation to interleukin-4 and −13 inhibition, and other corticosteroid-effects were not assessed [41].

The small number of participants enrolled in this study was the main limitation, especially as five of the 17 recruited could not manage or tolerate the techniques required for PExA or FeNO. These relatively high rates of failure are likely due to the severity of breathlessness experienced in some of our patients with severe asthma. Moreover, the first tests of all exhaled measurements were intended to be performed before ICS use at the baseline visit; however, this was not achievable due conflicts with patient's clinic schedule. Another potential limitation was noted from the repeated measurements performed over a narrow interval (2 h). PExA test in particular requires high effort over a sustained period of time. Although the study has demonstrated a pattern change in some of VOC profiles, the test was not evaluated in a placebo group to guarantee the change was due to ICS use.

In summary, this study demonstrates that levels of FeNO, exhaled VOC and EBT all change after ICS use. Furthermore, collected PEx mass is associated with lung function and the PExA test is potentially useful in clinical settings. Future, larger, studies are required to validate and reproduce these findings, and may provide novel clinically useful non-invasive markers for corticosteroid adherence, sensitivity, and efficacy.

Acknowledgments

Fahad Alahmadi was supported by Taibah University, Madinah, Saudi Arabia. M Wilkinson was funded by an EPSRC iCASE PhD studentship (EP/M507490/1). S J Fowler is supported by the NIHR Manchester Biomedical Research Centre. We acknowledge the support of Manchester University Foundation Trust severe asthma team and the NIHR Manchester Clinical Research Facility in carrying out this study.

Data availability statement

The data generated and/or analysed during the current study are not publicly available for legal/ethical reasons but are available from the corresponding author on reasonable request.

Short- and medium-term effect of inhaled corticosteroids on exhaled breath biomarkers in severe asthma

Article metrics

Submit

Author e-mails

Author affiliations

Author notes

ORCID iDs

Dates

Peer review information

Abstract

1. Introduction