METHODS

Between September 1999 and November 2001 143 consecutive preschool children (age ⩽7 years) referred to the unit of clinical physiology in the Division of Allergy, Helsinki University Central Hospital for the measurement of lung function (impulse oscillometry) and exhaled nitric oxide were studied. One hundred and thirty seven children (96%) had satisfactory IOS measurements and 102 (71%) performed acceptable FE_NO measurements. The proportion of successful measurements was strongly associated with the age of the children, and the age of the youngest children who performed the FE_NO measurement successfully was 3.8 years. Only those children who performed both tests satisfactorily were considered eligible for the study. In addition, the measurements from two children were discarded because of equipment failure and those from four children were discarded due to a history of chronic lung disease of prematurity.

The remaining 96 children were divided into three groups according to their history: The first group consisted of children with previously diagnosed asthma on regular medication—that is, treated asthma (n=25). The diagnosis had been based on typical clinical history, symptoms, and signs according to consensus statements.¹² The second group (n=21) consisted of children with persistent or recurrent respiratory symptoms associated with recent (within the previous 3 months) wheezing relieved by β₂ agonist therapy. Clinical examination and chest radiographs were performed to exclude acute infections and rare causes of wheeze before the lung function tests were performed. Based on clinical judgement, bronchoscopy was not considered to be indicated in any of these children with late onset wheeze. We considered that these children had probable asthma.¹² The third clinical group (n=46) consisted of children with chronic (persistent or recurrent) cough only, without wheezing episodes. The symptoms had lasted at least 6 weeks and acute or chronic respiratory infections were excluded, based on clinical examination, chest or sinus radiographs, before lung function tests were performed. The children in the probable asthma and chronic cough groups were referred for the first time because of a suspicion of asthma and had not used any anti-inflammatory medication (corticosteroids, cromones or antileukotriene drugs) for at least 2 months before the study. All the children in the group with treated asthma were on inhaled steroids at the time of the study. The demographic characteristics of the study groups are shown in table 1.

Control subjects were chosen from a sample of 62 age matched (4.0–7.0 years) healthy non-atopic children attending kindergartens¹³ who had satisfactory FE_NO and IOS measurements. They did not have any present or chronic respiratory symptoms, asthma or atopic disease, and their skin prick tests for common respiratory allergens were negative. With regard to demographic data, there were no significant differences between the study groups and healthy controls.

The definition of atopy in this study was based on skin prick testing using 10 common inhalant allergens (SQ, ALK, Horsholm, Denmark). The reaction was regarded as positive if the wheal diameter was 3 mm or more and the control solutions gave expected results. At the time of testing none of the children had experienced a respiratory tract infection in the preceding 2 weeks or showed signs of clinical obstruction (wheeze or shortness of breath). Short acting β₂ agonists were withheld for at least 12 hours preceding the test. The study was approved by the institutional ethics committees of Helsinki University Hospital and Espoo Social and Health Care Center, and written informed consent was obtained from the parents of all participating children.

Exhaled nitric oxide was measured using a chemiluminescence analyser (CLD 77 AM, Eco Physics, Duernten, Switzerland) connected to a computerised system (Exhaled Breath Analyzer, Aerocrine AB, Stockholm, Sweden) and calibrated with a certified NO calibration gas mixture (AGA Gas BV, Amsterdam, Netherlands). The standard single exhalation technique recommended by the American Thoracic Society,¹⁴ subsequently adopted also for children,¹⁰ was applied. Children were seated without a nose clip and were asked to fill their lungs completely with NO-free air, and thereafter to exhale slowly against a calibrated resistor of 200 cm H₂O/l/s (Hans Rudolf Inc, Kansas City, MO, USA) with a mean flow of approximately 50 ml/s for at least 6 seconds. The flow was measured with a heated pneumotachograph (Hans Rudolph Inc). A variation of 40–60 ml/s in mean and instantaneous exhaled flow was allowed. Measurements were repeated until 2–3 exhalations with specified flows and acceptable plateaus were obtained and the FE_NO values agreed within 10% or within 5% of their mean value. The mean value of these measurements was recorded as the final result.

Lung function was measured by impulse oscillometry (IOS; Jaeger, Würzburg, Germany). The method and equipment have been previously described in detail.^6,13 The output pressure and flow signals were analysed for their amplitude and phase difference to determine the resistance (Rrs) and reactance (Xrs) of the respiratory system, components of the respiratory impedance (Zrs). The pneumotachograph of the device was calibrated daily and the system was also checked against a reference impedance of 0.2 kPa/l.s. During the measurement the child was in a sitting position, breathing quietly through a mouthpiece. A nose clip was used and the cheeks were supported by the hands of the investigator. Measurements were repeated in order to obtain three acceptable data sets which were used to calculate the mean value for Rrs and Xrs at 5 Hz. After the baseline measurements, the children received salbutamol in a dose of 0.3 mg administered via a Babyhaler. The lung function measurements were repeated 15 minutes after inhalation to assess the bronchodilator response. In the group of healthy controls, 49 children took part in the bronchodilator test.

Linear regression methods were applied to the results in healthy children to create prediction intervals for the log transformed data of oscillometric parameters and FE_NO. Standing and sitting height, age, weight, body surface area (BSA), and sex were tested as possible predictors using stepwise regression analysis. Standing height was found to be the best independent variable for both tests. Sex was not a significant predictor in any of the measured variables. In patient groups the deviation from the predicted value was expressed as multiples of the residual standard deviation.¹⁵ The bronchodilator responses in Rrs5 and Xrs5 were expressed as the nominal change (post – prebronchodilator value, ΔRrs5 and ΔXrs5) or as the percentage change compared with the predicted baseline value (ΔRrs5%pred and ΔXrs5%pred).

The results in the clinical groups were compared with those of healthy controls using ANOVA and by calculating the mean differences with 95% confidence interval (CI). For post hoc comparisons of ANOVA, Fisher’s PLSD test was used. The discriminative usefulness of baseline lung function, bronchodilator responses, and FE_NO was evaluated and compared by constructing ROC curves¹⁶ where sensitivity versus 1 – specificity was plotted for each possible cut off level. For this analysis, the group with probable asthma was labelled as diseased compared with the group of healthy controls. For each variable the area under the curve (AUC) with 95% confidence interval was determined. From each ROC curve we determined the ideal cut off level which corresponds to the closest point to the top left hand corner and which discriminates between the absence or presence of disease most efficiently. The respective sensitivity, specificity, and predictive values were compared.

RESULTS

The concentration of FE_NO, respiratory resistance, and reactance measured with IOS were all significantly related to age and height of the healthy children. Standing height was the best independent variable, and introducing other factors (sitting height, age, weight or BSA) did not significantly improve the coefficient of determination of the predicted equations. With increasing height FE_NO increased slightly (r=0.29; p=0.02), Rrs5 decreased (r=0.50; p<0.0001), and Xrs5 increased (r=0.56; p<0.0001). In fig 1 FE_NO is shown as a function of standing height in healthy children with regression line and 95% prediction interval. Details of the regression equations for IOS variables have been described elsewhere.¹³ The bronchodilator response expressed as the nominal change (ΔRrs5 and ΔXrs5) was significantly related to age, height, and baseline value, but in terms of percentage change compared with the predicted values (ΔRrs5%pred and ΔXrs5%pred) independent of these factors.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Exhaled nitric oxide concentration (FE_NO) as a function of standing height in healthy non-atopic children. The solid line represents the line of regression (mean (SE) Ln (FE_NO, ppb) = 2.67 (1.15) × Ln (Standing height, cm) − 11.15 (5.48); r² = 0.08) and the dashed lines show the 95% prediction interval (RSD=0.504).

Baseline lung function, bronchodilator responses, and FE_NO are shown in table 2. Children in the probable asthma group had the highest concentration of NO in exhaled air with a mean difference from healthy controls of 16.8 ppb (95% CI 12.0 to 21.5). Baseline lung function in this group was also significantly different from healthy controls, characterised by increased Rrs5 (mean difference 0.13 kPa/l.s, 95% CI 0.03 to 0.22) and decreased Xrs5 (mean difference −0.08 kPa/l.s, 95% CI −0.13 to −0.03), and the bronchodilator responses were significantly greater. FE_NO values in children in the treated asthma group were not significantly different from those in healthy children. However, they had significant changes in baseline lung function. In children with chronic cough FE_NO values were significantly higher than in healthy children (mean difference 3.9 ppb, 95% CI 0.8 to 7.0) but their baseline lung function was normal. The bronchodilator responses in the chronic cough and treated asthma groups were not significantly different from those in healthy children.

In the chronic cough group the mean (SE) concentration of FE_NO in atopic children was 14.0 (3.0) ppb which was significantly higher than in non-atopic children (mean difference 8.5 ppb, 95% CI 2.1 to 14.9). In children with probable or treated asthma FE_NO was not significantly associated with atopy. Compared with atopic children with probable asthma with mean (SE) FE_NO of 24.5 (3.7) ppb, atopic children with chronic cough (mean difference 10.5 ppb, 95% CI 0.7 to 21.7) and treated asthma (mean difference 15.9 ppb, 95% CI 6.3 to 25.6) had significantly lower FE_NO.

Boxplot presentations of Rrs5, Xrs5, ΔRrs5%pred, and FE_NO in the study groups are shown in fig 2. In the combined clinical series of patients (n=96) the concentration of FE_NO was related to baseline Xrs5 (r=0.20, p=0.04) and to bronchodilator responses expressed as ΔRrs5 (r=0.31, p=0.002) and ΔRrs5%pred (r=0.29, p=0.004) but not significantly to Rrs5, ΔXrs5, or ΔXrs5%pred. Within the separate patient groups the correlations were not significant.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Boxplot presentation of the distribution of respiratory resistance and reactance at 5 Hz (Rrs5 and Xrs5), bronchodilator responsiveness (ΔRrs5%pred), and exhaled nitric oxide (FE_NO) in the study groups.

Based on the distribution of the healthy controls, the 5th (for Xrs5, ΔRs5%pred and ΔXrs5%pred) and 95th (for FE_NO and Rrs5) percentiles were determined and the numbers of abnormal test results in the study groups were calculated. In the probable asthma group 17 children (81%) had abnormal FE_NO concentrations but Rrs5 and Xrs5 were abnormal in only five (24%) and 11 (52%), respectively. Abnormal reversibility (ΔRrs5%pred or ΔXrs5%pred) was present in six children (29%) with probable asthma. The results of the discriminative properties of FE_NO and IOS variables are shown by ROC analysis in fig 3. FE_NO had the best discriminative capacity (AUC 0.91, 95% CI 0.83 to 0.96) followed by the lung function indices Rrs5 (AUC 0.77, 95% CI 0.67 to 0.86) and ΔRrs5%pred (AUC 0.76, 95% CI 0.64 to 0.85; fig 3). The optimal cut off level for FE_NO was 1.5 SD above predicted, corresponding approximately to a value of 9.7 ppb giving a sensitivity of 86% and specificity of 92% (table 3). In particular, FE_NO had a high negative predictive value (95%).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

ROC characteristics of FE_NO, baseline lung function (Rrs5, Xrs5), and bronchodilator responsiveness (ΔRrs5%pred and ΔXrs5%pred) in discriminating between children with probable asthma (n=21) and healthy controls (n=62).

DISCUSSION

Exhaled nitric oxide as an inflammatory marker of asthma was superior to measures of lung function assessed by the oscillometric technique in distinguishing preschool children with probable asthma. The diagnostic value of FE_NO^17–19 and the oscillometric method ^5,8 have been previously studied in preschool children with asthma. We compared the usefulness of these new techniques in our series of children at the age where the diagnosis of asthma has traditionally been largely based on clinical history. With the assumption that FE_NO reflects airway eosinophilia in asthma, the results emphasise the presence of airway inflammation at the early stages of asthma, even in preschool children.

Exhaled nitric oxide has been proposed as a non-invasive inflammatory marker of the airways. High concentrations have been found particularly in children with atopic asthma,²⁰ while the role of FE_NO in reflecting airway inflammation of non-atopic asthma is still unclear. Non-atopic wheeze is common in early childhood, but the proportion of IgE associated wheeze becomes more dominant in new cases at the age group of the present study.²¹ This was reflected in our consecutive series of children with probable asthma who were mostly atopic. Furthermore, the institution where the study was performed specialises in allergic disorders which may have resulted in selection bias for the referring stage in favour of atopic children. The small number of non-atopic children with asthma did not allow subanalyses of ROC in the present study, so we were unable to determine the discriminative capacity of FE_NO in non-atopic asthma. Atopy alone does not seem to explain the variation in FE_NO since the atopic children with chronic cough had significantly lower values than atopic children with probable asthma.

In children with asthma FE_NO has been found to correlate with sputum eosinophilia⁹ as well as eosinophilia in biopsy specimens,²² and it is also associated with other clinical characteristics of asthma such as airway hyperresponsiveness,²³ bronchodilator responses,²⁴ and exercise induced bronchoconstriction.²⁵ In our patients FE_NO was significantly associated with baseline lung function and with bronchodilator responses, although within the groups the correlations were not statistically significant. This loose association probably reflects the fact that lung function changes and airway inflammation are different aspects of asthma and do not necessarily co-exist in individual children.

Only a few reports of FE_NO in preschool asthmatic or healthy children are available.^17,19,26 In particular, studies based on the single exhalation technique recently recommended by the American Thoracic Society are lacking.^10,14 In these guidelines children are expected to exhale against a resistor for at least 6 seconds with a constant flow of approximately 50 ml/s. Although a slightly larger variation in exhaled flow (40–60 ml/s) was allowed in the present study than is recommended, the applicability of this standard online exhalation method was restricted to children aged ⩾4 years because of problems with cooperation which is a limitation of the technique. Only 71% of the children aged 4–7 years were capable of performing acceptable measurements. The mean FE_NO observed in the healthy children studied (5.3 ppb) was slightly less than that reported by Baraldi et al¹⁸ in healthy children aged 4–13 years using a special flow driven method (9.6 ppb), and slightly higher than that reported by Buchvald and Bisgaard¹⁹ in healthy children aged 2–5 years using controlled tidal breathing at a fixed flow rate (3 ppb). The observed age and height dependency of FE_NO in the present study is in agreement with the findings of Franklin et al.²⁷

The oscillometric technique has been found to be a useful method for assessing lung function and bronchodilator responsiveness in young children.^5,6,8,28 This technique was chosen for the present study because it is particularly suitable for young children as it requires less cooperation than spirometry and can be performed in most preschool children. This was confirmed by the high success rate (96%) of IOS measurements in the present study. Previously, Hellinckx et al⁴ were unable to find differences between healthy children and those with stable asthma when baseline lung function and bronchodilator responses were investigated with IOS. In contrast, we in the present study and Nielsen and Bisgaard⁸ have found significant differences between similar groups. This is probably explained by differences in selection criteria.

In young children there is no lung function method that would be regarded as the gold standard in the diagnosis of asthma. In the present study of preschool children we chose the history of recent recurrent wheeze relieved by β₂ agonists as an indication of probable asthma. Although in infancy wheeze may not accurately predict the development of asthma, recurrent wheezing in children of preschool to school age is highly predictive of asthma and constitutes the basis for the diagnosis in settings where other causes have been excluded and the presence of asthma is likely.¹² Therefore, in our children (age 4–7 years) with recurrent wheezing the diagnosis of asthma may be regarded as highly probable. We chose to use the group with probable asthma for discrimination analysis rather than the treated asthma group because these children were not on anti-inflammatory medication which could have confounded the results. Furthermore, the children in the probable asthma group had recent symptoms and were referred for the first time with suspected asthma. We therefore believe that they represent more closely the population in whom the diagnostic methodologies are applied and should be evaluated.

Wheeze is a sign of airflow limitation so it is not surprising that children in the probable asthma group had the highest degree of airway obstruction (increased resistance). As in the present study, preschool wheezing has been associated with increased responses to bronchodilators in earlier studies.²⁹ In preschool children wheeze is also associated with signs of airway inflammation expressed as high concentrations of NO in exhaled air.¹⁷

Our study is the first to combine lung function data with FE_NO values in a clinical group of preschool children. The most important finding was that, in the group of children with probable asthma characterised by recent wheeze, signs of airway inflammation (as expressed by FE_NO) were more constantly present than changes in lung function, and FE_NO could discriminate between these children and healthy children better than any of the lung function indices. In particular, bronchodilator responsiveness, considered one of the hallmarks of asthma diagnosis, was present less frequently than high levels of FE_NO. This is probably due to the variable nature of bronchial obstruction in asthma; in a cross sectional analysis such as the present study some asthmatic subjects are likely to have normal lung function. Alternatively, it may be that, in the early stages of the disease, signs of airway inflammation and symptoms precede those of abnormal lung function as we have shown in adults with asthma-like symptoms.¹¹

Persistent cough is one of the cardinal symptoms of asthma, but it may be associated with many other disorders. Some investigators argue that asthma is rarely the cause of persistent cough.³⁰ In most children with recurrent cough the symptoms are probably reflections of recurrent respiratory infections, although at the time of testing no acute infection was present. In some children chronic cough may have been a sign of cough variant asthma. We therefore consider that the aetiology of the symptoms in the group with chronic cough was heterogenous. In agreement with this, we found signs of airway inflammation and abnormal lung function only in a few of these children. In our series the lung function of children with chronic cough was not significantly altered, although some authors have reported increased bronchodilator responsiveness in patients with cough alone.²⁹ No reports on FE_NO in preschool children with persistent cough have previously been published. We found significantly raised FE_NO values in some of these children, suggesting airway inflammation. FE_NO in this group was significantly associated with atopy, assessed by skin prick tests. The young children with recurrent and persistent cough constitute a large but difficult diagnostic challenge. For these children, objective measures of lung function and airway inflammation are urgently needed for targeting anti-inflammatory therapy. Reports on adult patients with persistent cough suggest that FE_NO may be a promising tool in identifying subjects with asthma.³¹

We conclude that FE_NO is superior to baseline lung function measures or indices of bronchodilator responsiveness assessed by the oscillometric technique in identifying preschool children with predominantly atopic probable asthma. The discriminative properties of FE_NO in non-atopic preschool children with wheeze still need to be clarified. The results emphasise the presence of airway inflammation in the early stages of asthma, even in young children.