Inhaled and systemic corticosteroid response in severe asthma assessed by alveolar nitric oxide: a randomized crossover pilot study of add-on therapy

Abstract

AIMS

Alveolar nitric oxide (CA_NO) is a potential biomarker of small airway inflammation. We investigated effects on CA_NO of the addition of coarse and fine particle inhaled corticosteroids to standard therapy in severe asthma.

METHODS

Severe asthmatics taking ≥1600 µg day⁻¹ budesonide or equivalent performed a randomized open-label crossover study. Subjects with FEV₁ < 80%, gas trapping and CA_NO≥2 ppb entered a 6 week dose-ramp run-in of fluticasone/salmeterol(FPSM) 250/50 µg twice daily for 3 weeks, then 500/50 µg twice daily for 3 weeks. Patients then received additional HFA-beclomethasone diproprionate (BDP) 200 µg twice daily or FP 250 µg twice daily for 3 weeks in a crossover. Participants then received prednisolone(PRED) 25 mg day⁻¹ for 1 week. Nitric oxide, lung function, mannitol challenge, systemic inflammatory markers and urinary cortisol were measured.

RESULTS

Fifteen completed per protocol: mean (SD) age 51 (12) years, FEV₁ 58 (13)% predicted, residual volume 193 (100)% predicted and mannitol_PD10 177 (2.8) µg. There was no significant difference between FPSM and add-on therapy for CA_NO. FPSM/BDP and FPSM/PRED suppressed broncial flux (Jaw_NO) and FE_NO compared with FPSM alone, but there was no significant difference between FPSM/BDP and FPSM/FP. ECP, e-selectin and ICAM-1 were suppressed by FPSM/PRED compared with FPSM and FPSM/FP but not FPSM/BDP. Plasma cortisol was significantly suppressed by FPSM/PRED.

CONCLUSION

In severe asthma, CA_NO is insensitive to changes in dose and delivery of inhaled corticosteroids and is not suppressed by systemic corticosteroids. Additional inhaled HFA-BDP reduced FE_NO and Jaw_NO without adrenal suppression. There was a trend to reduction in FE_NO and Jaw_NO with additional FP but this did not reach statistical significance. PRED reduced FE_NO and Jaw_NO with suppression of systemic inflammatory markers and urinary cortisol.

WHAT IS ALREADY KNOWN ABOUT THIS SUBJECT

• Extra fine inhaled corticosteroid formulations such as HFA beclomethasone may be used to target the small airways in asthma. However, it is unclear which outcome measures may detect such small airways effects in patients with severe asthma.

WHAT THIS STUDY ADDS

• The alveolar fraction of exhaled nitric oxide is not a sensitive outcome measure for following effects of extra fine particle inhaled beclomethasone in severe asthma. However there were significant effects on nitric oxide fractions measured at 50 ml s⁻¹ and bronchial flux with the extra fine formulation, which did not produce significant cortisol suppression.

Introduction

Asthma is characterized by inflammation of the airways associated with smooth muscle hypertrophy, bronchial hyper-reactivity and airway remodelling [1, 2]. Whilst inhaled corticosteroids (ICS) are effective for the treatment for most asthmatics, 5–15% of patients are poorly controlled despite combination therapy, or require high dose therapy to control their disease [3]. Pathology series suggest that in severe asthma, disease may be more distally distributed compared with mild-to-moderate disease [2, 4, 5].

Fractional exhaled nitric oxide (NO) measured at a flow rate of 50 ml s⁻¹ (FE_NO) is correlated with airway eosinophila and has been used as a surrogate of steroid responsiveness [6]. In recent years extended exhaled NO techniques have been developed which calculate values representative of NO derived from the large conducting airways and the small airways and alveoli, respectively, the so-called ‘two compartment model’ of NO production [7]. The technique involves the measurement of exhaled NO at several flow rates from which alveolar NO (CA_NO) and bronchial flux (Jaw_NO) can be derived. The majority of traditional FE_NO is derived from Jaw_NO[8].

Early papers reported that CA_NO was elevated in severe asthmatics suggesting that it could be a useful biomarker for the monitoring of small airway disease in severe asthma [9, 10]. It is now recognized that the two compartment model was an over-simplification in that it does not account for NO diffusion between compartments. Hence a proportion of NO produced proximally (Jaw_NO) will diffuse back into the alveolar compartment resulting in false elevation of CA_NO and vice versa. Mathematical adjustments are now used to correct for this [11, 12]. This correction is especially important when Jaw_NO (and hence FE_NO) are elevated as is often the case in severe asthma. Hence in severe asthma, the response of CA_NO to different preparations of corticosteroids has not been studied in a controlled trial with correction for back-diffusion.

The aim of the present study was to investigate the effects on CA_NO of the addition of coarse or fine particle ICS to standard therapy in severe asthma, using oral prednisolone (PRED) as a positive control. Other surrogates of airway and systemic inflammation were assessed for comparison.

Methods

Study design

An open-label randomized crossover study was performed (Figure 1). Participants were never-smokers or ex-smokers with <10 pack years, aged 16–65 years, a respiratory physician diagnosis of asthma, prescribed ≥1000 µg day⁻¹ fluticasone or 1600 µg day⁻¹ budesonide in the previous year. Subjects required evidence of persistent airflow obstruction (post-bronchodilator FEV₁ : FVC ratio < 0.7, FEV₁ < 80%) and gas trapping (total lung capacity and residual volume > 100%). Exclusion criteria included diagnosis of bronchiectasis or allergic broncho-pulmonary aspergillosis and long term oral corticosteroid use. Subjects were also required to have an elevated uncorrected alveolar NO (≥2 ppb) following a 6 week dose-ramp run-in of fluticasone/salmeterol (FPSM) 500 µg day⁻¹ for 3 weeks, then 1000 µg day⁻¹ for 3 weeks via dry powder inhaler (Seretide Accuhaler, Allen & Hanbury).

Figure 1.

Study diagram

For the treatment periods, participants continued FPSM 1000 µg day⁻¹ and were additionally randomized to extra-fine HFA-beclomethasone diproprionate (BDP) 400 µg day⁻¹ (BDP) (Qvar easibreathe, Teva Pharmaceuticals) or fluticasone propionate 500 µg day⁻¹ (FP) (Flixotide Accuhaler, Allen & Hanbury) for 3 weeks in a crossover fashion. Following a 3 week run-out all participants were allocated PRED 25 mg day⁻¹ (PRED) for 1 week. Inhaler technique was assessed and corrected at every visit using an In-check dial (Clement Clarke International Limited, Essex, UK). Compliance was re-enforced at each visit and dose counters checked for DPI treatments. Computer generated simple randomization was performed to allocate subjects to cross-over sequence.

At baseline and following each treatment the following measurements were performed: extended exhaled NO, spirometry, impulse oscillometry, body plethysmography, asthma quality of life questionnaire (AQLQ), asthma control questionnaire (ACQ), mannitol challenge and blood for ECP, ICAM-1 and E-selectin. Subjects took study medication on the morning of visits, but withheld ‘as required’ medications. Leukotriene receptor antagonists were not permitted during the study. Alveolar NO was the primary outcome and other measurements secondary. All measurements were performed at the Asthma and Allergy Research Group at Ninewells Hospital, Dundee and Perth Royal Infirmary, Perth. The study was approved by the Tayside Committee on Medical Research Ethics and registered on clinicaltrials.gov (NCT00829257).

Measurements

Lung function

Spirometry was performed in accordance with guidelines [13] using a SuperSpiro spirometer (Micro Medical Ltd, Chatham, Kent, UK). A Jaeger Masterscreen (Erich Jaeger, Hoechberg, Germany) was used for impulse oscillometry (IOS) and full body plethysmography. IOS was performed as previously published [14]. If FEV₁ > 60% the provocative dose of mannitol required for a 10% fall in FEV₁ (mannitol PD₁₀) was performed as previously published [15].

Extended exhaled nitric oxide

An on-line chemiluminescence analyzer (Niox, Aerocrine AB, Solna, Sweden) was used to quantify exhaled NO at multiple flow rates, 50 ml s⁻¹, 100 ml s⁻¹ and 200 ml s⁻¹. Online graphs of NO exhalation manoeuvres were examined to ensure that an adequate plateau had been achieved. Measurements were repeated as necessary to ensure a minimum of two satisfactory readings at each flow rate with <10% variation. CA_NO and Jaw_NO were derived using the two-compartment models described by Tsoukias & George and Tsoukias et al. [7, 16] with correction for axial diffusion as described by Kerckx et al. [12].

Laboratory analysis

Urine was collected overnight for 10 h prior to study visits. Urinary creatinine was determined using a Roche Cobas analyzer (Roche, Burgess Hill, UK; coefficient of variance (CV) 4.9%) and cortisol by radioimmunoassay (Diasorin, Bracknell, UK; CV 8.6%). ECP was determined using the UNICAP method (Phadia Ltd, Milton Keynes, UK; CV 4.0%). E-selectin and ICAM-1 were measured through use of enzyme immunoassays (Diaclone, Besancon, France; CV 1.5% and 2.8%).

Statistical analysis

Data were assessed for normality using the Shapiro-Wilk test and inspection of boxplots. Non-Gaussian data were log-transformed prior to analysis. Values for baseline (FPSM), treatment limbs (FPSM/FP and FPSM/FP) and positive control limb (FPSM/PRED) were compared using an anova of repeated measures followed by Bonferroni correction for multiple comparisons. A priori power calculations predicted that 14 participants in crossover design ensured 90% power, with an α error of 0.05, to detect a 0.5 ppb change in CA_NO, assuming a within patient standard deviation of 0.37 ppb based on previous studies in asthmatics [9, 17]. Analyses were performed using SPSS version 17.0 (Chicago, Illinois, USA) or GraphPad PRISM version 5.01 (San Diego, California, USA).

Results

Fifty-two screens were performed of which 26 were suitable to enter the 6 week dose ramp run-in. Following run-in 17 met entry criteria for randomization to treatments. Two subjects were withdrawn post randomization due to exacerbation requiring systemic steroids (one on each treatment) (see figure 2). Demographic data and medication at screen are shown in Table 1. Mean (SD) age and FEV₁ were 51 (12) years and 58 (13)% predicted respectively. All were current non-smokers and 13/15 were never smokers.

Figure 2.

Consort diagram

Table 1.

Demographics at screen

Screen number	Age(years)/Gender	BDP (µg )	LABA (L), Nasal steroid (N), LAMA(T), Antihistamine (A)	FEV1 (%)	Residual volume (%)	Resistance at 5 Hz (R5%)	Mannitol PD10
1	50/M	1600	L	77	159	202	95.53
2	52/F	2000	L	66	456	115	30.39
6	52/M	1000	LNA	67	83	109	139.73
3	63/M	1600	LN	54	142	215	170.37
10	37 M	2000	LA	71	170	129	13.28
23	65F	1000	L	49	216	192	337.32
27	48 M	2000	L	67	130	130	192.50
31	40/M	2000	LA	44	224	213	NR
32	28/M	1000	L	64	140	214	96.94
36	43 M	2000	L	70	162	286	455.89
37	65/M	1600	–	69	147	151	129.27
38	60/M	2000	–	57	128	153	443.57
39	61/F	2000	N	48	185	142	NR
40	52/M	2000	L	63	153	200	121.69
41	52/M	2000	LN	38	188	306	NR
Mean (SD)	51 (10)	1635 (448)		58 (13)	193 (100)	184 (56)	177 (2.81)

Challenge not performed as FEV₁ < 50% predicted. BDP, beclomethasone diproprionate equivalent; LABA, long acting beta-agonist; LAMA, long acting anti-muscarinic.

There was no significant difference between FPSM and any add-on therapy for corrected CA_NO. Mean (95% CI) values were 1.39 ppb (0.48, 2.31) for FPSM, 1.79 ppb (0.58, 3.00) for FPSM/FP, 1.71 ppb (0.69, 2.73) for FPSM/BDP and 1.29 ppb (0.62, 1.96) for FPSM/PRED (Figure 3, Table 2). The FPSM/BDP and FPSM/PRED suppressed Jaw_NO and FE_NO compared with FPSM alone (Figure 3, Table 2, Table 4). For FE_NO values were 22.5 ppb (17.0, 27.9) and 21.4 ppb (16.4, 26.4) for FPSM/BDP and FPSM/PRED, respectively compared with 30.7 ppb (21.1, 40.4) for FPSM alone. There was a numerical trend for a reduction in FE_NO and Jaw_NO with FPSM/FP, but this did not reach statistical significance (P= 0.09 for both). There was no significant difference between FPSM/FP and FPSM/BDP (Figure 3, Table 2, Table 4).

Figure 3.

Extended nitric oxide measurements stratified by treatment group. Data displayed as mean with 95% confidence intervals. *P < 0.05 compared with FPSM. FP, fluticasone; FPSM, fluticasone/salmeterol; BDP, beclomethasone diproprionate; PRED, prednisolone

Table 2.

Inflammation and adrenal axis

Treatment	FPSM	FPSM/FP	FPSM/BDP	FPSM/PRED
Exhaled nitric oxide
FENO (ppb)	30.7 (21.1–40.4)	24.3 (18.9–29.7)	22.5 (17.0–27.9)*	21.4 (16.4–26.4)*
JawNO (nL.s−1)	2667 (1744–3590)	2008 (1517–2500)	1902 (1370–2434)*	1795 (1323–2266)*
CANO (ppb)	1.39 (0.48–2.31)	1.79 (0.58–3.00)	1.71 (0.69–2.73)	1.29 (0.62–1.96)
Uncorrected CANO (ppb)	3.74 (3.17–4.31)	3.59 (2.44–4.74)	3.37 (2.40–4.34)	2.90 (2.26–3.54)
Plasma ECP	33.3 (17.6 to 49.0)	31.2 (21.2 to 41.7)	23.1 (14.1 to 32.1)	15.9 (12.0 to 19.7)**
Plasma E-selectin	95.3 (82.2 to 108.3	103.7 (87.8 to 119.5)	94.0 (77.4 to 110.7)	78.0 (64.3 to 91.7) **
Plasma ICAM-1	546 (485 to 607)	540 (475 to 605)	513 (456 to 571)	479 (418 to 540) **
Mannitol PD10 (µg)†	135 (57–320)	133 (60–294)	182 (72–459)	119 (55–260)
Adrenal axis
08.00 h plasma cortisol†	394 (338 to 459)	419 (349 to 503)	407 (356 to 466)	61 (31 to 121)***

^*P < 0.05 compared with FPSM.

^**P < 0.05 compared with FPSM and FPSM/FP.

^***P < 0.05 compared with FPSM, FPSM/FP and FPSM/BDP.

^†Presented as geometric mean and 95% CI.

FP, fluticasone; FPSM, fluticasone/salmeterol; BDP, beclomethasone diproprionate; PRED, prednisolone.

Table 4.

Difference in means and 95% confidence intervals for extended nitric oxide measurements

		Mean difference	95% CI	Significance
FENO (ppb)
FPSM	vs. FPSM/FP	7.85	(–1.10, 16.80)	0.094
	vs. FPSM/BDP	10.41	(1.46, 19.36)	0.048
	vs. FPSM/PRED	11.18	(2.23, 20.13)	0.041
FPSM/FP	vs. FPSM/BDP	2.57	(–6.38, 11.52)	0.485
	vs. FPSM/PRED	3.33	(–5.62, 12.28)	0.379
FPSM/BDP	vs. FPSM/PRED	0.76	(–8.19, 9.71)	0.829
JawNO (nl s−1)
FPSM	vs. FPSM/FP	765.1	(–103.5, 1634)	0.093
	vs. FPSM/BDP	932.6	(63.96, 1801)	0.049
	vs. FPSM/PRED	1027	(158.6, 1896)	0.047
FPSM/FP	vs. FPSM/BDP	167.5	(–701.1, 1036)	0.630
	vs. FPSM/PRED	262.1	(–606.5, 1131)	0.465
FPSM/BDP	vs. FPSM/PRED	94.63	(–774.0, 963.3)	0.783
CANO (ppb)
FPSM	vs. FPSM/FP	–0.56	(–2.35, 1.24)	0.459
	vs. FPSM/BDP	–0.32	(–2.12, 1.47)	0.651
	vs. FPSM/PRED	–0.15	(–1.94, 1.64)	0.832
FPSM/FP	vs. FPSM/BDP	0.23	(–1.56, 2.03)	0.741
	vs. FPSM/PRED	0.41	(–1.39, 2.20)	0.631
FPSM/BDP	vs. FPSM/PRED	0.17	(–1.62, 1.97)	0.806
Uncorrected CANO (ppb)
FPSM	vs. FPSM/FP	0.20	(–1.33, 1.73)	0.740
	vs. FPSM/BDP	0.60	(–0.92, 2.13)	0.355
	vs. FPSM/PRED	0.87	(–0.66, 2.40)	0.213
FPSM/FP	vs. FPSM/BDP	0.40	(–1.13, 1.93)	0.520
	vs. FPSM/PRED	0.67	(–0.86, 2.19)	0.312
FPSM/BDP	vs. FPSM/PRED	0.27	(–1.26, 1.79)	0.660

FP, fluticasone; FPSM, fluticasone/salmeterol; BDP, beclomethasone diproprionate; PRED, prednisolone.

There was no statistically significant difference between treatments for any measure of pulmonary function including spirometry, IOS, body plethysmography and bronchial challenge, although there was a numerical trend for improvement in body plethysmography with FPSM/PRED (Tables 2 and 3).

Table 3.

Lung function

Treatment	FPSM	FPSM/FP	FPSM/BDP	FPSM/PRED
Spirometry
FEV1 (l)	2.16 (1.81 to 2.51)	2.18 (1.82 to 2.54)	2.21 (1.81 to 2.61)	2.26 (1.89 to 2.63)
FEF25–75 (l)	1.35 (0.98 to 1.71)	1.29 (0.92 to 1.66)	1.36 (0.95 to 1.77)	1.38 (1.00 to 1.76)
FVC (l)	3.42 (3.03 to 3.81)	3.40 (3.00 to 3.81)	3.45 (3.03 to 3.87)	3.49 (3.08 to 3.90)
RVC (l)	3.70 (3.25 to 4.16)	3.49 (3.03 to 3.94}	3.51 (3.06 to 3.96)	3.66 (3.26 to 4.06)
FEV1 (%)	63.4 (57.6 to 69.1)	64.2 (59.0 to 69.4)	64.6 (58.5 to 70.7)	65.5 (59.6 to 71.4)
FEF25–75 (%)	34.8 (26.3 to 43.3)	34.0 (25.4 to 42.5)	35.0 (25.7 to 44.3)	35.6 (26.6 to 44.6)
FVC (%)	83.2 (78.5 to 88.0)	83.0 (80.4 to 85.6)	83.9 (81.2 to 86.7)	83.6 (78.5 to 88.7)
RVC (%0	88.0 (83.5 to 92.5)	82.5 (77.9 to 87.0)	82.6 (79.2 to 86.0)	84.8 (81.3 to 88.4)
Impulse oscillometry
R5 (kPa l−1 s)	0.47 (0.39 to 0.56)	0.45 (0.40 to 0.59)	0.44 (0.41 to 0.66)	0.47 (0.37 to 0.68
R5-R20	0.13 (0.23 to 0.13)	0.11 (0.06 to 0.23)	0.11 (0.06 to 0.25)	0.09 (0.03 to 0.21)
X5 (kPa l−1 s)	−0.15 (−0.23 to −0.13)	−0.17 (−0.24 to −0.13)	−0.15 (−0.26 to −0.12)	−0.15 (−0.33 to −0.12)
Body plethysmography
sRaw (kPa s−1)	1.87 (1.31 to 2.42)	1.94 (1.43 to 2.45)	2.02 (1.51 to 2.54)	1.57 (1.02 to 2.13)
sRaw (%)	166 (114 to 218)	175 (127 to 222)	180 (131 to 229)	142 (93 to 192)
RV (l)	3.28 (2.58 to 3.97)	3.16 (2.80 to 3.51)	3.44 (2.60 to 4.28)	2.76 (2.42 to 3.10)
RV (%)	159 (113 to 205)	144 (127 to 161)	170 (113 to 227)	130 (117 to 143)
RV/TLC	0.44 (0.38 to 0.51)	0.46 (0.41 to 0.50)	0.46 (0.41 to 0.52)	0.41 (0.36 to 0.46)

Presented as mean and (95% CI) except impulse oscillometry which is median (IQR). % shown as percent predicted for race, age and gender. FP, fluticasone; FPSM, fluticasone/salmeterol; BDP, beclomethasone diproprionate; PRED, prednisolone.

Plasma ECP, e-selectin and ICAM-1 were significantly suppressed by FPSM/PRED compared wit FPSM and FPSM/FP but not FPSM/BDP. However FPSM/BDP was not significantly reduced compared with FPSM/FP or FPSM (Figure 4, Table 2).

Figure 4.

Plasma ECP, E-selectin and cortisol stratified by treatment group. Data displayed as mean and 95% confidence intervals with exception of plasma cortisol, displayed as geometric mean and 95% confidence intervals. **P < 0.05 compared with FPSM and FPSM/FP, ***P < 0.05 compared with FPSM, FPSM/ FP and FPSM/BDP. FP, fluticasone; FPSM, fluticasone/salmeterol; BDP, beclomethasone diproprionate; PRED, prednisolone

Plasma cortisol was significantly suppressed by FPSM/PRED compared with all other limbs (Figure 4, Table 2). There was no difference in overnight urinary cortisol : creatinine ratio (OUCC) for inhaled treatment (FPSM, FPSM/FP or FPSM/BDP). OUCC was not tested in the FPSM/PRED limb due to cross-reactivity of the assay to urinary metabolites.

ACQ and AQLQ values are shown (Figure 5). There was no significant difference in symptoms scores for any treatment.

Figure 5.

Asthma control questionnaire (ACQ) and asthma quality of life questionnaire (AQLQ) stratified by treatment group. Data displayed as mean and 95% confidence intervals. FP, fluticasone; FPSM, fluticasone/salmeterol; BDP, beclomethasone diproprionate; PRED, prednisolone

Discussion

This is the first study to examine the potential role of CA_NO for assessing therapies in severe asthma. Whilst early studies suggested CA_NO could be an attractive biomarker to assess the small airways we have demonstrated it is insufficiently sensitive to detect a dose response to inhaled corticosteroids or a response to systemic therapy.

Secondary outcomes in the present study showed that inhaled HFA-BDP produced significant reductions in FE_NO and Jaw_NO when added to a maximal dose of FPSM, without causing adrenal suppression. Similar effects on FE_NO and Jaw_NO were observed with oral PRED, but were associated with significant cortisol suppression. The addition of high dose FP showed a numerical trend for reduction which did not reach statistical significance (Table 2, Table 4). There was no statistically significant difference between FPSM/BDP and FPSM/FP.

‘Severe’ or ‘refractory’ asthma is difficult to characterize for research purposes and recommendations have been reviewed several times in recent years [3, 18–21]. The joint American Thoracic Society statement on refractory asthma suggests that criteria should include measures of airway obstruction and level of therapy a patient requires for stability [3]. We therefore selected participants on the basis of lung function as well as the dose of inhaled corticosteroids. In order to include a positive control limb of systemic steroids and to select a group stable enough to tolerate changes to medication we did not include subjects on long term oral corticosteroid therapy. In this regard our patients may not be considered as severe as cohorts in other studies on refractory asthma [22–24] but nonetheless represent a refractory group as evidenced by a lack of improvement in lung function and symptoms even with systemic therapy. To ensure enrolment of subjects with airflow obstruction due to asthma, we sought evidence of gas trapping with body plethysmography (mean residual volume 193% predicted), excluded current smokers or ex-smokers with >10 pack year history and confirmed bronchial hyper-reactivity in all subjects with sufficient respiratory reserve to tolerate a bronchial challenge (mean mannitol PD₁₀ 177 µg).

In order to select participants with evidence of small airways inflammation and room for further suppression of CA_NO, entry criteria also included a CA_NO > 2 ppb following the 6 week run-in period. This level was selected on the basis of pilot data, but at the time of protocol design the importance of correction for axial diffusion was not widely recognized and therefore an uncorrected value was used [25].

Whilst several studies have reported elevated levels of CA_NO in severe asthma, only one other study has assessed its responsiveness to corticosteroids [9, 10, 17, 25]. A number of studies have been performed in milder asthma with regards to suppression of CA_NO using ICS. However methods and results are varied and results have been conflicting [26]. In an uncontrolled study by Berry et al. CA_NO was suppressed in severe asthmatics by oral PRED but not by a doubling of inhaled FP [9]. The authors concluded that systemic treatment may be required to target adequately the small airways, in keeping with Bel et al. who have shown benefits of parenteral steroids in refractory asthma [23]. However, in the paper by Berry et al. [9], no correction for axial diffusion was applied and further inspection of the data reveals that FE_NO was significantly suppressed with PRED (32.4 vs. 17.8 ppb) but not by a doubling of ICS (41.6 vs. 41.3 ppb). Hence the apparent fall in uncorrected CA_NO may represent a significant reduction in back-diffusion rather than a true reduction in alveolar NO. In the present study there was no significant suppression of uncorrected values, but the respective changes in FE_NO and Jaw_NO were far more modest (Table 2). Our findings are also consistent with a more recent observational study of 15 asthmatics demonstrating that CA_NO is not elevated during acute exacerbations and change in FE_NO output was due to flux rather than CA_NO[27]. This paper also suggested that after correction for Jaw_NO, CA_NO was not elevated in asthma compared with age matched controls, in keeping with the lack of response in the present paper and pilot work by our department [25].

There are a number of possible reasons that CA_NO was not suppressed in the present study. The run-in of FPSM may have maximally suppressed CA_NO preventing any further suppression. The run-in period included a dose-ramp from FPSM 500 µg day⁻¹ to FPSM 1000 µg day⁻¹ to assess if a dose–response to CA_NO could be demonstrated. As in the main study, this showed a dose–response in FE_NO and Jaw_NO but not corrected CA_NO (data not shown). It has been shown that for FE_NO there is a level of maximal suppression for an individual which varies dependent on factors such as age, gender, atopy and height [28]. Similarly, it is plausible there is a level of maximal suppression for CA_NO. If so, inclusion of individuals with a CA_NO >2 ppb on FPSM 1000 µg day⁻¹ may have inadvertently selected individuals maximally suppressed at a high level, rather than allowing further ‘room for improvement’ as initially hypothesized. Finally it is possible that the steroids simply have no significant effect on CA_NO in severe asthma, as has been suggested in COPD [29].

It has been suggested that small airway disease may be an important target for asthma therapy. With the development of extra-fine particle ICS it is now possible to achieve greater small airways drug deposition [30]. This is particularly pertinent in individuals with airflow obstruction as this leads to more central distribution of inhaled therapies [31]. We had hypothesized that extra-fine HFA-BDP may achieve additional suppression of CA_NO compared with a biological equivalent dose of coarse particle ICS. Whilst this was not observed, extra-fine BDP achieved significant suppression of FE_NO and Jaw_NO compared with FPSM without causing adrenal suppression. Similarly whilst FPSM/PRED significantly suppressed systemic markers of asthmatic inflammation (ECP, e-selectin and ICAM-1) compared with FPSM and FPSM/FP, they were not significantly lower than extra-fine HFA-BDP (Table 2). The consistency of this trend across ECP, e-selectin and ICAM-1 would suggest that a larger study powered for these markers may have demonstrated significant suppression with FPSM/BDP compared with FPSM and FPSM/FP, though this cannot be confirmed with the present data. Alternatively, it is possible that systemic markers of inflammation can only be suppressed with appreciable systemic bioavailability, which was demonstrated for PRED but not inhaled therapies. In this regard only moderate doses of HFA-BDP were used as add-on therapy. Hence suppression of inflammatory markers and cortisol may have been observed at higher doses. Whilst the combined doses of inhaled therapy in the study would be very high for most asthmatics, it should be recognized that severe airflow obstruction reduces the lung bioavailability of inhaled steroid as has been shown previously with FP [32–35].

Although HFA-BDP showed additional suppression of FE_NO and flux, the clinical significance of such changes in steroid treated asthmatics is not established. Smith et al. have examined ‘personal best FE_NO’ in 73 non-severe asthmatics comparing FE_NO at a threshold where control is lost, ‘optimized’ inhaled FP and oral corticosteroid. They also found levels were significantly lower on oral steroid compared with optimized inhaled FP, but had no extra-fine particle comparator [28]. Further adequately powered studies would be required to establish whether extra-fine particle ICS could achieve near ‘personal best’ FE_NO and whether this would confer greater asthma control with long term therapy.

We were also interested to see that there was a non-significant numerical trend towards improvement in mannitol airway hyperreactivity with FPSM/BDP, though this was less than the biological variability of one doubling dilution. Again, larger studies would be warranted to assess the relative benefits of different particle size over a wider dose range and sufficiently powered to assess effects on mannitol challenge, exacerbations and systemic inflammatory markers.

For lung function, whilst there was a numerical trend for improvement with FPSM/PRED, there were no statistically significant changes. These were secondary outcomes and as such the study was not sufficiently powered to detect small changes. Furthermore, we felt it unethical to use our standard medication withholding time of three half-lives because of our cohorts relative instability. Instead we asked participants to use their study medication on the morning of the visit and as such measurements of lung function, IOS and body plethysmography represent post bronchodilator measurements. Any additional bronchodilator effect from corticosteroids is therefore likely to be small and difficult to detect even with techniques sensitive to changes in airway obstruction undetectable by spirometry alone [36]. Similarly, no significant improvement was seen in symptom scores. We would interpret the lack of improvement in symptoms and lung function, even on oral therapy, as good evidence of the refractory nature of our cohort's asthma. Lung function and symptoms were secondary outcomes and as such the study was not powered for these outcomes. Large long term studies would be required to confirm or refute whether changes in NO would translate to clinically meaningful patient outcomes.

We must acknowledge that our study had no placebo control arm which is always desirable in clinical trial design. The severity of our patients meant that a true placebo arm with no therapy was not possible. Instead a ‘positive’ control arm was used to characterize maximal suppression. We believe this is still a clinically meaningful comparison in individuals who, by definition, have treatment resistant disease. We acknowledge that our cohort was older than a general population of severe asthmatics, perhaps reflecting a population willing and able to engage in active clinical research. We acknowledge several further limitations to our study design which were required to minimize risks of exacerbation in a severe cohort: a baseline with a full washout would have been potentially hazardous, full bronchial challenge in all patients would have been dangerous due to low FEV₁, as would inhalation of hypertonic saline for induced sputum. The stringency of the inclusion criteria meant that numbers were necessarily limited, though a cross-over design was employed to maximize statistical power. Finally, some may feel that our study design was contrived as addition of high dose ICS to combination therapy is not a standard approach in refractory disease and is not included in guidelines. Whilst we acknowledge this, we feel there is a paucity of data to inform on a pharmacological approach to refractory asthma and much of the published work focuses on non-pharmacological and co-morbid aspects of disease. We are aware anecdotally of pulmonologists who may indeed add high dose ICS to combination therapy rather than step-up to maintenance dose oral corticosteroid. Therefore whilst the study design is artificial it is nonetheless of interest pharmacologically.

In conclusion, in severe asthma, inhaled HFA-BDP produced additional reductions in FE_NO and Jaw_NO without adrenal suppression. Oral PRED produced similar effects but with additional suppression of systemic inflammatory markers and cortisol. There was a numerical trend for reduction with FP but this did not reach statistical significance and no significant difference between FPSM/BDP and FPSM/FP. Alveolar NO is insensitive to changes in dose and delivery of ICS and is not suppressed by systemic corticosteroids. This suggests that CA_NO is either an insensitive test of reversible small airway disease, or that this cohort of patients require non-steroidal immunosuppressive treatments. CA_NO is therefore unlikely to be a useful tool in measuring response to treatments targeted at small airway disease in refractory asthma.

Competing Interests

BJL has received unrestricted educational grant support from TEVA for the present study and has also received funds for giving post graduate educational talks and attending advisory boards for TEVA. There are no other competing interests to declare.

Source of funding

Investigator led study supported by an unrestricted educational grant from TEVA Pharmaceuticals Inc. USA.

Contributions

Conception and design: PAW, SV, BJL. Analysis and interpretation: PAW, PMS, BJL. Drafting of manuscript for important intellectual content: PAW, PMS, SV, BJL.