Airway hyperresponsiveness to mannitol and methacholineand exhaled nitric oxide: A random-sample population study

Asger Sverrild, MD, Celeste Porsbjerg, MD, PhD, Simon Francis Thomsen, MD, PhD, and

Vibeke Backer, MD, DMSc Copenhagen, Denmark

Abbreviations used

AHR: Airway hyperresponsiveness

BPT: Bronchial provocation test

eNO: Exhaled nitric oxide

IQR: Interquartile range

RDR: Response-dose ratio

ROC: Receiver operating characteristic

Background: Studies of selected patient groups have shown thatairway hyperresponsiveness (AHR) to mannitol is more specificthan methacholine for the diagnosis of asthma, as well as morecloselyassociatedwithmarkers of airway inflammation inasthma.Objective: We sought to compare AHR to mannitol andmethacholine and exhaled nitric oxide (eNO) levels in anonselected population sample.Methods: In 238 young adults randomly drawn from thenationwide civil registration list in Copenhagen, Denmark,AHR to mannitol and methacholine, as well as levels of eNO,were determined, and the association with asthma was analyzed.Results: In diagnosing asthma the specificity of methacholineand mannitol was 80.2% (95% CI, 77.1% to 82.9%) and 98.4%(95% CI, 96.2% to 99.4%), respectively, with a positivepredictive value of 48.6% versus 90.4%, whereas the sensitivitywas 68.6% (95% CI, 57.1% to 78.4%) and 58.8% (95% CI,50.7% to 62.6%), respectively. In asthmatic subjects AHR tomannitol was associated with increased eNO levels (positiveAHR to mannitol: median, 47 ppb [interquartile range, 35-68ppb]; negative AHR to mannitol: median, 19 ppb [interquartilerange, 13-30 ppb]; P 5 .001), whereas this was not the case forAHR to methacholine (median of 37 ppb [interquartile range,26-51 ppb] vs 24 ppb [interquartile range, 15-39 ppb], P 5 .13).Conclusion: In this random population sample, AHR tomannitol was less sensitive but more specific than methacholinein the diagnosis of asthma. Furthermore, AHR to mannitol wasmore closely associated with ongoing airway inflammation interms of increased eNO levels. (J Allergy Clin Immunol2010;126:952-8.)

Key words: Asthma, airway hyperresponsiveness, mannitol, metha-choline, diagnosis, exhaled nitric oxide, inflammation, epidemiology,population

From the Respiratory Research Unit, Department of Respiratory Medicine, Bispebjerg

University Hospital.

The Danish Agency for Science, Technology and Innovation, an institution under the

Danish Ministry of Science, Technology and Innovation, supported the study with a

1-year scholarship. Pharmaxis provided the research team with an unrestricted grant,

with which one of the research assistants was employed. Moreover, mannitol test kits

were provided by Pharmaxis Ltd (Frenchs Forest, NSW, Australia).

Disclosure of potential conflict of interest: C. Porsbjerg receives honoraria from

Pharmaxis and receives research funding for the Danish Agency of Science and

Technology. The rest of the authors have declared that they have no conflict of interest.

Received for publication December 3, 2009; revised August 18, 2010; accepted for pub-

lication August 19, 2010.

Available online October 13, 2010.

Reprint requests: Celeste Porsbjerg, MD, PhD, Respiratory Research Unit, Department

of Respiratory Medicine, Bispebjerg Hospital, Bispebjerg Bakke 23, 2400 Copenha-

gen NV, Denmark. E-mail: Porsbjerg@dadlnet.dk.

0091-6749/$36.00

� 2010 American Academy of Allergy, Asthma & Immunology

doi:10.1016/j.jaci.2010.08.028

952

Bronchial provocation tests (BPTs) have a higher sensitivityfor the diagnosis of asthma than spirometry or reversibilitytesting,1,2 but BPTs have been somewhat limited by being moretechnically challenging, being more time-consuming, and requir-ing more equipment. Furthermore, the traditional direct BPTs us-ing histamine or methacholine have a relatively weak relationshipwith airway inflammation,3,4 which might account for their rela-tively low specificity, with false-positive test results in nonasth-matic subjects. Histamine and methacholine act directly on thesmooth muscle cells of the airways to cause bronchoconstriction,and airway hyperresponsiveness (AHR) to these agents might beseen in the absence of airway inflammation.3,4 Direct BPTs aretherefore less useful for confirming the presence of asthma withactive airway inflammation.5

In comparison, indirect BPTs, such as use of hypertonic salineand adenosine monophosphate, act through inducing release ofbronchoconstricting mediators, such as histamine, prostaglan-dins, and leukotrienes, from inflammatory cells in the airways,and these tests have been shown to reflect airway inflammationbetter than direct BPTs.4,5 However, practical issues, such as theneed for specific equipment, have limited the use of indirect tests.An indirect BPT using mannitol powder has been developed thatconsists of a simple single-use test kit that has the practical advan-tages of being easy to use, being safe, and requiring less equip-ment.6 We have previously shown a closer relationship betweenAHR to mannitol and markers of airway inflammation (sputumeosinophil percentage and exhaled nitric oxide [eNO] level) com-pared with AHR to methacholine in a selected group of asthmaticsubjects who were not receiving anti-inflammatory treatment.7

This is a random-sample population study of AHR to mannitoland methacholine in teenagers and young adults. Data on the di-agnostic properties of mannitol BPTs have recently been pub-lished,8 and the present article compares the diagnostic validityof the mannitol BPTand the methacholine BPTand their relation-ship with eNO level as a marker of airway inflammation in asth-matic and nonasthmatic subjects.

METHODS

Study designThe study is a cross-sectional population study performed at theRespiratory

Research Unit, Copenhagen University Hospital Bispebjerg, Denmark. The

J ALLERGY CLIN IMMUNOL

VOLUME 126, NUMBER 5

SVERRILD ET AL 953

same data were used in the recently published characterization of the

diagnostic properties of inhaled mannitol in asthmatic subjects.8 A sample of

1000 young adults between the ages of 14 and 24 years was randomly drawn

from the civil registration list. All subjects received a validated self-

administered asthma and rhinitis screening questionnaire with 20 questions

adopted from the American College of Allergy, Asthma, and Immunology

screening program extendedwith questions concerning tobacco consumption.9

All participants attended 1 study visit, which included (in order of

execution) measurement of eNO at a rate of 50 mL/min, lung function

measurement, a skin prick test, a BPT with inhaled mannitol, a BPT with

methacholine, and a reversibility test to an inhaled b2-agonist. The results of

the methacholine test have been shown not to be influenced by the perfor-

mance of a previousmannitol test,10 and this test was performedwhen the sub-

jects had reached a level of 95% of their baseline FEV1.

In line with international guidelines on bronchial provocation, all partic-

ipants were told to withhold use of antiasthma drugs before the examination.11

This included use of inhaled corticosteroids within 12 hours before the visit.

All participants completed a semistructured interview at the visit, including

questions regarding respiratory symptoms, such as chest tightness, cough,

wheezing, exercise-induced dyspnea, nocturnal symptoms, and rhinitis; atopic

dermatitis; and familial predisposition to asthma, allergy, or both. A single in-

vestigator carried out all tests, measurements, and interviews. A group of non-

responders were contacted by telephone to test differences between

responders and nonresponders. Further details on the study population,

methods, and guidelines have previously been published.8

SpirometrySpirometry was performed according to European Respiratory Society

recommendations.12 FEV1 and forced vital capacity weremeasured with a 7-L

dry wedge spirometer (Vitalograph, Buckingham, United Kingdom) cali-

brated weekly.

Mannitol BPTDry-powder mannitol (Aridol; Pharmaxis LTd, Frenchs Forest, NSW

Australia) was administered according to the recommendations of the

manufacturer, and FEV1 was recorded in line with current guidelines.12

FEV1 recorded after inhalation of a 0-mg placebo capsule constituted baseline

lung function. The challenge was stopped at a decrease in FEV1 of 15% or

greater from baseline values or when the maximum cumulative dose of 635

mg had been administered. A positive challenge response was defined as a de-

crease in FEV1 of at least 15% after inhalation of 635 mg of mannitol or less.

Methacholine challengeBronchial provocationwithmethacholine up to a cumulative dose of 8mmol

was performed with the Spira dosimeter (Spira Respiratory Care Center Ltd,

Hameenlinna, Finland) by using the dosimetric method previously described by

Yan et al.13 A positive challenge response was defined as a decrease in FEV1 of

at least 20% after inhalation of 8 mmol of methacholine or less.

The response-dose ratio (RDR) values for methacholine and mannitol were

calculated as the percentage decrease in FEV1 after the last dose divided by the

cumulative dose in micromoles or milligrams.

eNOeNO levels were measured online (rate, 0.05 L/s) with the Nitric Oxide

Analyzer (NIOX; Aerocrine AB, Solna, Sweden) and according to American

Thoracic Society guidelines.14

In the overall assessment of whether the participant had asthma, a cutoff

of 30 ppb was used (see the section on the diagnosis of asthma for more

details). However, for the analysis of the relation between airway inflam-

mation and AHR to mannitol and methacholine, a cutoff of 26 ppb for

eNO was chosen because this level has been shown to have the highest

sensitivity and specificity for predicting a sputum eosinophil percentage of

greater than 3%.15,16

Skin prick testsA skin prick test to 10 aeroallergens (birch [Betula species], grass [Phleum

pratense]mugwort, horse, dog, cat [Felis domesticus], house dust mite [Der p

1 and Der f 2], and fungi [Alternaria andCladosporium species; ALK-Abello,

Hørsholm, Denmark]) was performed according to the European Academy of

Allergy and Clinical Immunology’s recommendations.17 Allergic sensitiza-

tion was defined as a positive skin prick test response to at least 1 of these

10 aeroallergens.

Asthma definitionA respiratory specialist performed the evaluation to ensure consistent

classification of the asthma diagnosis.8 Interpretation of lung function tests

and other paraclinical data followed current international guidelines.18

A diagnosis of asthma was consistent with symptoms of asthma within the

last 12 months in combination with either a eNO level of greater than 30

ppb, a history of allergic rhinoconjunctivitis, dermatitis, a positive skin prick

test response, a familial predisposition to atopic disease, nonallergic rhinocon-

junctivitis, or an FEV1/forced vital capacity ratio of less than 75%. The exam-

iner was blinded to the results of the mannitol and methacholine tests but had

otherwise free access to the abovementioned clinical information. Subjects

with clinical remission of disease for longer than 12 months were classified

as having no asthma. The clinical diagnosis of asthmawas not objectively con-

firmed because the results from the bronchial provocation challenges consti-

tuted the primary end point of the analysis.

Statistical analysisData were analyzed with SPSS version 17.0 (SPSS, Inc, Chicago, Ill). Data

are reported as means (SDs) for normally distributed variables and as medians

(interquartile ranges [IQRs]) for nonnormally distributed variables. For

analysis of parametric data, x2 tests, 2-sample t tests, and ANOVA were

used, and the Fisher exact test, the Kruskal-Wallis test, and the Mann-

Whitney U test were used for analysis of nonparametric data.

A receiver operating characteristic (ROC) curve was constructed, plot-

ting RDRs to mannitol and methacholine against the diagnosis of asthma.

The overall accuracy of the test was measured as the area under the ROC

curve.

RDR values for mannitol and methacholine, as well as eNO levels, were

nonnormally distributed when assessed with Kolmogorov-Smirnoff analysis

and were therefore log transformed. The correlation between the log RDR to

mannitol and that to methacholine was analyzed with Pearson correlation

analysis. After log transformation, log eNO values were still not normally

distributed (Kolmogorov-Smirnoff 5 0.04), and the Spearman correlation

coefficient was therefore used to assess the correlation between the degree of

airway responsiveness to mannitol and methacholine and the level of eNO.

The proportion of subjects with increased eNO levels (>26 ppb) among

asthmatic subjects with AHR to mannitol versus AHR to methacholine was

compared by calculating the z value as follows:

z5ðORmannitol� ORmethacholineÞ2=ð½SEmannitol�21½SEmethacholine�2Þ � 1;

with OR defined as the odds ratio.

To assess whether the relationship between AHR to mannitol and

methacholine and eNO level was independent of other factors that might

normally influence the level of eNO, logistic regression analysis was

performed in the 51 asthmatic subjects, including factors that are known to

potentially influence eNO levels, such as age, height, sex, smoking, use of

inhaled steroids, atopy, rhinitis, and AHR to mannitol and methacholine, as

independent variables and an eNO level of greater than 26 ppb as the

dependent variable. Because of the limited number of subjects, the enter

method was used, whereby all variables are entered into the model simulta-

neously. Because there was a significant association between the response

to mannitol and methacholine, 2 separate regression analyses were per-

formed, one including AHR to mannitol and one including AHR to

methacholine.

TABLE I. Baseline characteristics

Current asthma (n 5 51) No asthma (n 5 187) P value

Age (y), (median [minimum-maximum]) 18 (15-24) 19 (14-24) .77

Sex (% female) 61% (31) 60% (113) .87

Atopy 77% (39) 32% (60) <.0001

Smoking (current) 29% 24% .32

FEV1% predicted, mean (95% CI) 92% (89% to 95%) 94% (92% to 95%) .32

FEV1/FVC ratio, mean (95% CI) 0.85 (0.82-0.87) 0.88 (0.87-0.89) .001

Use of ICS 16% 0% <.0001

FVC, Forced vital capacity; ICS, inhaled corticosteroid.

TABLE II. Results of 238 randomly selected adolescents tested

with inhaled mannitol and methacholine: 51 asthmatic subjects

and 187 nonasthmatic subjects

TABLE III. Diagnostic properties of inhaled mannitol and meth-

acholine in 238 randomly selected subjects

Sensitivity Specificity PPV NPV

Methacholine 69 (57-78) 80 (77-83) 49 (40-56) 90 (87-93)

Mannitol 59 (51-63) 98 (96-99) 91 (78-97) 90 (88-91)

Numbers are percentages (95% CIs). A positive mannitol test response was defined as

a decrease in FEV1 of 15% or greater at a cumulative dose of 635 mg or less, whereas

a positive methacholine test response was defined as a decrease in FEV1 of 20% or

greater at a cumulative dose of 8 mmol or less. The mannitol data have already been

published by Sverrild et al.8

J ALLERGY CLIN IMMUNOL

NOVEMBER 2010

954 SVERRILD ET AL

RESULTS

Population characteristicsPopulation characteristics are shown in Table I. Most asthmatic

subjects had normal or near-normal lung function in terms ofFEV1 and FEV1/forced vital capacity ratio, although the percentpredicted FEV1 was slightly lower compared with that seen innonasthmatic subjects. Most were nonsmokers, and the preva-lence of smoking was comparable between asthmatic and non-asthmatic subjects. Atopy was observed in 77% of asthmaticsubjects compared with 32% of nonasthmatic subjects.Current asthma was diagnosed in 51 (21%) of the 238 subjects

participating in the study, most of whom had normal lung function(FEV1). A further 12 subjects had previously had symptoms ofasthma but had not experienced any symptoms within the past 12months. Inhaled steroids were used by 16% at the time of the study,and an additional 31% (16/51) had previously used inhaled ste-roids, although only 1 had used them within the last 12 months.

Diagnostic properties of mannitol versus

methacholineThe results of the 2 tests in all 238 subjects are shown in Tables

II and III. As recently published,7 the sensitivity and specificity ofmannitol were 59% (95% CI, 51% to 63%) and 98% (95% CI,96% to 99%), respectively, for a diagnosis of asthma, and the pos-itive predictive value (PPV) and negative predictive value (NPV)were 91% (95% CI, 78% to 97%) and 90% (95% CI, 88% to91%), respectively. In comparison, methacholine, using 8 mmol

as the cutoff, had a sensitivity and specificity of 69% (95% CI,57% to 78%) and 80% (95% CI, 77% to 83%), respectively, fora diagnosis of asthma, corresponding to a PPVof 49% (95% CI,40% to 56%) and an NPVof 90% (95%CI, 87% to 93%). The dif-ferences in specificity and predictive values reflect that amongnonasthmatic subjects, 37 (19.8%) responded to methacholine,whereas only 3 (1.6%) responded to mannitol.In asthmatic subjects the median RDRs for mannitol and

methacholine were 0.026 (IQR, 0.014-0.094) and 10.8 (IQR, 1.8-61.7), respectively, whereas in nonasthmatic subjects the medianRDRs for mannitol and methacholine were 0.005 (IQR, 0.002-0.009) and 1.0 (IQR, 0.54-2.08), respectively.Fig 1 shows an ROC curve of RDRs to methacholine and man-

nitol versus the diagnosis of asthma. The area under the curve ofmethacholine amounts to 84.9% (95% CI, 79.1% to 90.8%), andthe area under the curve for mannitol to 89.1% (95% CI, 83.2% to95.0%). Decreasing the cutoff of a positive methacholine test re-sponse to 1 mmol increased the specificity to 97.9% (95% CI,95.6% to 99.1%) but with a simultaneous decrease in sensitivityto 43.1% (95% CI, 34.8% to 47.8%).Of the 51 subjects with asthma, 37 had experienced symptoms

within the last 4 weeks. Using a 4-week cutoff instead of 12months to define asthma, the sensitivity and specificity ofmethacholine were 70.3% and 77.1%, respectively, whereas thesensitivity and specificity of mannitol were 56.8% and 94.0%,respectively.

Relationship between AHR to mannitol and

methacholineAmong asthmatic subjects, 26 responded to both mannitol and

methacholine, 4 responded only to mannitol, and 9 respondedonly to methacholine. The Pearson correlation coefficient for logRDR mannitol and log RDR methacholine in asthmatic subjectswas 0.73 (P < .001, Fig 2).

FIG 1. ROC curve of the degree of AHR tomannitol andmethacholine (RDR)

for predicting a diagnosis of asthma.

FIG 2. Scatter plot of log RDR to mannitol versus log RDR to methacholine

in asthmatic subjects.

J ALLERGY CLIN IMMUNOL

VOLUME 126, NUMBER 5

SVERRILD ET AL 955

Using inhaled mannitol to predict the outcome of a methacho-line challenge, the sensitivity was reduced to 38.9% (95% CI,32.3% to 42.7%), and the specificity was 97.0% (95% CI, 94.1%to 98.6%). Furthermore, PPV and NPV were 84.8% (95% CI,70.6% to 93.2%) and 78.5% (95% CI, 76.2% to 79.9%),respectively.

AHR and eNO levels in subjects with current

asthmaThe association between the degree of hyperresponsiveness

and eNO levels was similar for mannitol and methacholine,although slightly stronger for mannitol (Spearman rho: log RDRmethacholine, 0.43; log RDR mannitol, 0.48; Fig 3).

However, a positive response to the mannitol challenge testmore clearly distinguished between asthmatic subjects with andwithout increased eNO levels. An eNO level of greater than 26ppb was found in 70% of asthmatic subjects with AHR tomannitol compared with 57% of subjects with AHR to methacho-line (P5 .007, Table II). Furthermore, subjects with AHR toman-nitol had significantly higher levels of eNO than subjects withoutAHR to mannitol (median eNO level, 47 ppb [IQR, 35-62 ppb] vs19 ppb [IQR, 13-30 ppb]), whereas the level of eNO did not differbetween subjects with and without AHR tomethacholine (medianeNO level, 37 ppb [IQR, 26-51 ppb] vs 24 ppb [IQR, 15-39 ppb],P 5 .13; Table IV).

AHR and eNO levels in nonasthmatic subjectsOnly 3 (1.6%) nonasthmatic subjects responded to mannitol;

all 3 had an eNO level of greater than 26 ppb, and 2 had previouslyhad asthma (Table IV). In the 37 (20%) nonasthmatic subjectswith AHR to methacholine, eNO levels were slightly higherthan in nonasthmatic subjects without AHR tomethacholine (me-dian eNO level, 17 ppb vs 13 ppb; P5 .03; Table IV). Of these 37

subjects, 6 had previously had asthma, and 3 of them had an eNOlevel of greater than 26 ppb. Another 2 subjects had eNO levelsgreater than 26 ppb but no history of asthma.

Regression analysisTo control for the effect of other factors potentially influencing

the level of eNO, a logistic regression analysis, including data onsex, age, height, smoking, allergic sensitization, rhinitis, andsteroid use, was performed, which showed that AHR to mannitolpredicted an eNO level of greater than 26 ppb independently ofthese factors (Table V). A similar logistic regression analysisincluding methacholine instead of mannitol showed that thiswas not the case for methacholine (Table V).

DISCUSSIONThis study compares the diagnostic properties of inhaled

mannitol and methacholine in a random population sample andthe correlation with the inflammatory marker eNO. We found thatmannitolwas significantlymore specific in the diagnosis of asthmathanmethacholine because 98%of the subjectswithout asthmahada negative mannitol test response compared with 80% when usingmethacholine, and the respective PPVs were 91% for mannitolcompared with 49% for methacholine. As expected based onprevious comparisons of direct and indirect BPTs, methacholinewasamore sensitive test because69%of the asthmatic subjects hada positive test result compared with 59% with mannitol. Impor-tantly, although decreasing the cutoff of methacholine from 8 to1mmol did increase the specificity of the methacholine test, it alsodecreased the sensitivity significantly. We have previously shownthat asymptomatic subjects with AHR to methacholine selectedfrom a random population study sample did not respond tomannitol.19 The present study confirms that mannitol is more spe-cific than methacholine in a random population sample.

FIG 3. Scatter plot of log eNO versus log RDR to mannitol (A) and log RDR

to methacholine (B) in asthmatic subjects.

J ALLERGY CLIN IMMUNOL

NOVEMBER 2010

956 SVERRILD ET AL

Of the 187 subjects without asthma, 37 (19.8%) had a positiveresponse to methacholine. The reasons other than asthma, whichcan explain AHR to methacholine, are many and well described(ie, technical, allergy, smoking, and normal variation). Back-ground rates for asymptomatic AHR tomethacholine are reportedfrom less than 10% to greater than 40%, depending on selectionand protocol.20 Female subjects are in general more responsivethan male subjects (possibly because of differences in airway cal-iber), and having 60% girls participating in the study, this mightbe a contributing factor to this outcome.However, it is also a well-described phenomenon that adoles-

cents experiencing clinical remission in their asthma remainhyperresponsive with increased eNO levels.21 Six of the 37 non-asthmatic subjects with AHR to methacholine had a history ofasthma, which was also the case for 2 of the 3 nonasthmatic

subjects with positive responses to mannitol. The definition ofasthma in this study was limited to those who had experiencedany symptomof asthmawithin the past 12months, and the subjectswith previous asthma symptoms and current AHR might havebeen in remission but with an increased risk of experiencing re-lapse of symptoms later in life. However, the majority of nonasth-matic subjects with AHR to methacholine denied having anycurrent or previous respiratory symptoms suggestive of asthma.Asthma, especially in young subjects, is often characterized by

an intermittent course of disease. Using the epidemiologicdefinition of asthma (12 months) holds the potential risk offalse-negative test results because of AHR returning to normal inperiods of clinical remission. However, the diagnostic propertiesof the 2 tests, when using a 4-week restriction in experiencedsymptoms to define asthma, showed no appreciable changescompared with a 12-month cutoff. Methacholine has for manyyears been thought of as a highly sensitive test in the diagnosis ofasthma. However, this is not supported by this study in that thesensitivity of methacholine was only slightly higher than that ofmannitol. Furthermore, in a recently published phase III study byAnderson et al,22 the sensitivity of methacholine was similar tothat of mannitol.There might be several explanations for this. First, methacho-

line might in fact not be a highly sensitive test in populationsamples. This is supported by another study using 8 mmol as thecutoff against a physician’s diagnosis of asthma, reporting an av-erage sensitivity of 56% and a specificity of 86%.23 Second, it hasbeen argued that the dosimeter method for administrating metha-choline shows a significantly reduced response in subjects withmild AHR compared with the tidal-breath method.24 A thirdpossibility is misclassification of disease, which is discussed infurther detail below.We found responsiveness to mannitol and methacholine to

correlate well, which tells us a high RDR to methacholine ingeneral can be expected in subjects with high RDRs to mannitol.This is also the case in more selected groups of asthmaticsubjects.8 That the 2 tests are correlating is not surprising becauseboth are thought to cover parts of the underlying hyperresponsive-ness. However, as suggested by the eNO findings in this study, it isunlikely the 2 tests cover the same elements of AHR and the un-derlying pathobiology. More defined subgrouping of asthmaticsubjects and in-depth investigation of underlying mechanisms(ie, inflammatory and genotypic features) are needed in this field.

Relationship between eNO level as a marker of

airway inflammation and AHRThere was a similar association between the level of eNO and

the degree of AHR to mannitol, as well as to methacholine, inasthmatic subjects, but a positive mannitol test response moreclearly differentiated between subjects with and without in-creased eNO levels. Furthermore, whereas very few nonasthmaticsubjects responded to mannitol, a significant number had AHR tomethacholine with or without increased eNO levels.The observation that the degree of responsiveness to both

mannitol and methacholine were related to eNO level mightreflect that although the indirect stimuli are dependent oninflammatory cells, such as eosinophils or mast cells, beingpresent and releasing sufficient amounts of bronchoconstrictingmediators, there is also an indirect association between theresponse to methacholine and ongoing airway inflammation in

TABLE IV. eNO levels in asthmatic and nonasthmatic subjects with and without AHR to mannitol and methacholine

Current asthma (n 5 51) No asthma (n 5 187)

Mannitol Methacholine Mannitol Methacholine

Negative

(n 5 21)

Positive

(n 5 30)

Negative

(n 5 16)

Positive

(n 5 35)

Negative

(n 5 184)

Positive

(n 5 3)

Negative

(n 5 150)

Positive

(n 5 37)

eNO (ppb), median (IQR) 19 (13-30) 47 (35-62) 24 (15-39) 37 (26-51) 14 (13-15) 46 (10-214) 13 (12-41) 17 (13-21)

P value .001 .13 <.0001 .03

eNO >26 ppb, %(n) 29 (6) 70 (21) 44 (7) 57 (20) 11 (21) 100 (3) 11 (16) 22 (8)

P value <.0001 .07 .01 .07

TABLE V. Determinants of eNO: Logistic regression analysis,

including AHR to mannitol and AHR to methacholine*

Mannitol Methacholine

OR 95% CI P value OR 95% CI P value

Age (y) 0.9 0.7-1.1 .19 0.9 0.7-1.1 .15

Sex (male 5 1) 2.5 0.3-17.7 .37 2.0 0.3-12.2 .47

Height (cm) 1.1 0.9-1.2 .24 1.1 0.9-1.2 .32

Current smoker 0.8 0.2-3.6 .75 0.8 0.2-3.4 .77

Current ICS use 4.9 0.5-50.0 .18 5.9 0.6-57.9 .13

Atopy 2.3 0.3-16.7 .41 2.6 0.4-15.3 .30

Rhinitis 2.1 0.1-33.3 .60 2.5 0.2-35.0 .51

AHR to mannitol 5.3 1.3-21.0 .02 — — —

AHR to methacholine — — — 1.8 0.4-7.3 .43

ICS, Inhaled corticosteroid; OR, odds ratio.

*Because of the association between the response to mannitol and methacholine, 2

separate regression analyses were performed, including RDRs to mannitol and

methacholine, respectively.

J ALLERGY CLIN IMMUNOL

VOLUME 126, NUMBER 5

SVERRILD ET AL 957

terms of increased smooth muscle responsiveness to methacho-line, as well as accessibility to the smooth muscle layer throughdisruption of the basal membrane.25

Approximately 30% of asthmatic subjects with a positiveresponse to mannitol had a normal eNO level and might not havehad airway eosinophilia. An explanation for this could be that theresponse to mannitol is mediated mainly through mast cells thatare responsible for the main release of bronchoconstrictingmediators, and a response to mannitol might therefore occur inthe absence of eosinophils or an increased eNO level.26,27

One third of asthmatic subjects with a negative mannitol testresponse had an eNO level of greater than 26 ppb. However, themedian eNO level of 19 ppb and upper IQR of 29 ppb in this groupindicates that these subjects generally had relatively low eNOlevels. Comparatively, the group that did respond to the mannitoltest had a median eNO level of approximately 47 ppb, which is theeNO level that has been shown to be optimal for predicting aresponse to steroid treatment.28 Clinically, the interpretation ofthese findings is that in someone with a positive mannitol testresponse, airway inflammation is likely to be present, whereas thesameassumptionmight not bebased ona response tomethacholine.

LimitationsIn lack of an operating gold standard of asthma, a clinical

diagnosis based on asthma symptoms, paraclinical data, atopicdisease, and familial predisposition was used. This strategy holdsseveral strengths, as well as weaknesses. Lack of accuracy is apotential bias in this study design because interobserver varia-bility cannot be assessed. On the other hand, the asthma diagnosisis consistent and can be reproduced.A participation rate of 25% puts the study at risk of selection

bias. However, a more detailed analysis of participants and acontrol group of nonresponders showed no differences in self-reported asthma, respiratory symptoms, and smoking, amongothers.8 Altogether, there is no strong evidence that the study sub-jects were not representative of the population as a whole.The study population was relatively young and in general had a

history of low tobacco consumption. This should be kept in mindwhen extrapolating these results to an older population whomighthave a longer history of asthma, a larger tobacco consumption,and other potential comorbidities, such as chronic obstructivepulmonary disease, which might all potentially have an effect onthe relationship between AHR to direct versus indirect BPTs andeNO levels because of airway remodeling, alteration of airwaycaliber, and elastic recoil, as well as an altered inflammatoryresponse in smokers.3,6

Regarding the use of eNO as a marker of airway inflammation,eNO levels have been shown to correlate with the degree of

airway eosinophilia in asthmatic subjects, the relationship is notlinear, and eosinophilia might be present in spite of a normaleNO levels and vice versa.16 However, any limitations of eNOlevel as a marker of inflammation would be expected to be sim-ilar in asthmatic subjects with AHR to mannitol or to methacho-line, and we believe it is unlikely that they have any significantinfluence on the present findings. Other factors might be associ-ated with increased levels of eNO, such as allergic sensitization,rhinitis, smoking, age, sex, and height, but after controlling forthese factors in a regression analysis, we found that there was stillan independent relationship between eNO level and AHR tomannitol.Taking these limitations into consideration, the present findings

support the concept that, when compared with methacholine, themannitol BPT is a more specific test for asthma with ongoingairway inflammation. This is also in keeping with previousobservations that responsiveness to mannitol decreases aftersteroid treatment, with an associated improvement in symptomscores,29,30 and thatAHR tomannitol predicts asthma deteriorationduring steroid dose downtitration.31 A practical advantage of themannitol test is the ease of performance of the test, which, com-bined with the high positive predictive value for the diagnosis ofasthma, might increase the use of BPTs in general.In conclusion, AHR to mannitol is more specific but less

sensitive than AHR to methacholine in the diagnosis of asthma inan unselected group of young adults. Furthermore, AHR tomannitol was associated with increased levels of eNO, whereassubjects without AHR to mannitol generally had normal eNOlevels. In comparison, eNO levels did not differ significantlybetween subjects with and without AHR to methacholine.

J ALLERGY CLIN IMMUNOL

NOVEMBER 2010

958 SVERRILD ET AL

Clinical implications: AHR to mannitol was less sensitive butmore specific than methacholine in the diagnosis of asthmaand more closely associated with increased eNO levels. Respon-siveness tomannitol might bemore reflective of asthmawith on-going airway inflammation.

REFERENCES

1. Crapo RO, Casaburi R, Coates AL, Enright PL, Hankinson JL, Irvin CG, et al.

Guidelines for methacholine and exercise challenge testing-1999. Am J Respir

Crit Care Med 2000;161:309-29.

2. Backer V, Groth S, Dirksen A, Bach-Mortensen N, Hansen KK, Laursen EM, et al.

Sensitivity and specificity of the histamine challenge test for the diagnosis of asthma

in an unselected sample of children and adolescents. Eur Respir J 1991;4:1093-100.

3. Henriksen AH, Lingaas-Holmen T, Sue-Chu M, Bjermer L. Combined use of ex-

haled nitric oxide and airway hyperresponsiveness in characterizing asthma in a

large population survey. Eur Respir J 2000;15:849-55.

4. Cockcroft DW, Davis BE. Diagnostic and therapeutic value of airway challenges in

asthma. Curr Allergy Asthma Rep 2009;9:247-53.

5. Van Den Berge M, Meijer RJ, Kerstjens HA, de Reus DM, Koeter GH, Kauffman

HF, et al. PC(20) adenosine 59-monophosphate is more closely associated with

airway inflammation in asthma than PC(20) methacholine. Am J Respir Crit

Care Med 2001;163:1546-50.

6. Anderson SD, Brannan J, Spring J, Spalding N, Rodwell LT, Chan K, et al. A new

method for bronchial-provocation testing in asthmatic subjects using a dry powder

of mannitol. Am J Respir Crit Care Med 1997;156:758-65.

7. Porsbjerg C, Brannan JD, Anderson SD, Backer V. Relationship between airway

responsiveness to mannitol and to methacholine and markers of airway inflamma-

tion, peak flow variability and quality of life in asthma patients. Clin Exp Allergy

2008;38:43-50.

8. Sverrild A, Porsbjerg C, Thomsen SF, Backer V. Diagnostic properties of inhaled

mannitol in the diagnosis of asthma: a population study. J Allergy Clin Immunol

2009;124:928-32, e1.

9. Winder JA, Nash K, Brunn JW. Validation of a life quality (LQ) test for asthma.

Ann Allergy Asthma Immunol 2000;85:467-72.

10. Gade EB, Thomsen SF, Porsbjerg C, Backer V. The bronchial response to mannitol

is attenuated by a previous methacholine test: but not vice versa. Clin Exp Allergy

2009;39:966-71.

11. Crapo RO, Casaburi R, Coates AL, Enright PL, Hankinson JL, Irvin CG, et al.

Guidelines for methacholine and exercise challenge testing-1999. This official

statement of the American Thoracic Society was adopted by the ATS Board of

Directors, July 1999. Am J Respir Crit Care Med 2000;161:309-29.

12. Miller MR, Hankinson J, Brusasco V, Burgos F, Casaburi R, Coates A, et al. Stand-

ardisation of spirometry. Eur Respir J 2005;26:319-38.

13. Yan K, Salome C, Woolcock AJ. Rapid method for measurement of bronchial

responsiveness. Thorax 1983;38:760-5.

14. ATS workshop proceedings: exhaled nitric oxide and nitric oxide oxidative metab-

olism in exhaled breath condensate: executive summary. Am J Respir Crit Care

Med 2006;173:811-3.

15. Shaw DE, Berry MA, Thomas M, Green RH, Brightling CE, Wardlaw AJ, et al.

The use of exhaled nitric oxide to guide asthma management: a randomized con-

trolled trial. Am J Respir Crit Care Med 2007;176:231-7.

16. Berry MA, Shaw DE, Green RH, Brightling CE, Wardlaw AJ, Pavord ID. The use

of exhaled nitric oxide concentration to identify eosinophilic airway inflammation:

an observational study in adults with asthma. Clin Exp Allergy 2005;35:1175-9.

17. Dreborg S. Skin tests used in type I allergy testing. Allergy 1989;44(suppl 10):

22-30.

18. Pellegrino R, Viegi G, Brusasco V, Crapo RO, Burgos F, Casaburi R, et al. Inter-

pretative strategies for lung function tests. Eur Respir J 2005;26:948-68.

19. Porsbjerg C, Rasmussen L, Thomsen SF, Brannan JD, Anderson SD, Backer V.

Response to mannitol in asymptomatic subjects with airway hyperresponsiveness

to methacholine. Clin Exp Allergy 2007;37:22-8.

20. Hewitt DJ. Interpretation of the "positive" methacholine challenge. Am J Ind Med

2008;51:769-81.

21. van Den Toorn LM, Prins JB, Overbeek SE, Hoogsteden HC, de Jongste JC. Adoles-

cents in clinical remission of atopic asthma have elevated exhaled nitric oxide levels

and bronchial hyperresponsiveness. Am J Respir Crit Care Med 2000;162:953-7.

22. Anderson SD, Charlton B, Weiler JM, Nichols S, Spector SL, Pearlman DS, et al.

Comparison of mannitol and methacholine to predict exercise-induced broncho-

constriction and a clinical diagnosis of asthma. Respir Res 2009;10:4.

23. James A, Ryan G. Testing airway hyperresponsiveness using inhaled methacholine

or histamine. Respirology 1997;2:97-105.

24. Cockcroft DW, Davis BE. The bronchoprotective effect of inhaling methacholine

by using total lung capacity inspirations has a marked influence on the interpreta-

tion of the test results. J Allergy Clin Immunol 2006;17:1244-8.

25. Moller GM, Overbeek SE, van Helden-Meeuwsen CG, Hoogsteden HC, Bogaard

JM. Eosinophils in the bronchial mucosa in relation to methacholine dose–response

curves in atopic asthma. J Appl Physiol 1999;86:1352-6.

26. Porsbjerg C, Lund TK, Pedersen L, Backer V. Inflammatory subtypes in asthma are

related to airway hyperresponsiveness to mannitol and exhaled NO. J Asthma

2009;46:606-12.

27. Brannan JD, Gulliksson M, Anderson SD, Chew N, Kumlin M. Evidence of mast

cell activation and leukotriene release after mannitol inhalation. Eur Respir J 2003;

22:491-6.

28. Smith AD, Cowan JO, Brassett KP, Filsell S, McLachlan C, Monti-Sheehan G,

et al. Exhaled nitric oxide: a predictor of steroid response. Am J Respir Crit

Care Med 2005;172:453-9.

29. Brannan JD, Koskela H, Anderson SD, Chan HK. Budesonide reduces sensitivity

and reactivity to inhaled mannitol in asthmatic subjects. Respirology 2002;7:37-44.

30. Koskela H, Hyvarinen L, Brannan JD, Chan H-K, Anderson SD. Sensitivity and

validity of three bronchial provocation tests to demonstrate the effect of inhaled

corticosteroids in asthma. Chest 2003;124:1341-9.

31. Leuppi JD, Salome CM, Jenkins CR, Anderson SD, Xuan W, Marks GB, et al.

Predictive markers of asthma exacerbations during stepwise dose-reduction of

inhaled corticosteroids. Am J Respir Crit Care Med 2001;163:406-12.