Interrupter resistance and oxygen saturation for methacholine challenge in young children

To the Editor:

In young children unable to perform reliable and reproducible spirometry, non-cooperative lung function techniques are necessary to measure bronchial hyperreactivity (BHR) during bronchial challenge [1]. Measuring the decrease in transcutaneous partial pressure of oxygen (P_tcO2) is a robust technique that detects increased ventilation–perfusion mismatch during bronchial challenge [2] in preschool and school-aged children [3–5], and a 20% decrease in P_tcO2 correlates to a 20% forced expiratory volume in 1 s (FEV₁) decrease in children aged 6–14 years [3] and in adults (with correlation to arterial oxygen tension) [6]. When neither spirometry nor P_tcO2 is available, other BHR outcomes can be measured such as wheezing that appears for mean±sd decreases of −44.7±14.5% in FEV₁ and −6.3±2.7% in transcutaneous saturation of oxygen (S_pO2) [7]. Respiratory resistances are easy to measure [8–10] but the relevant threshold for BHR is not yet defined and an at least 35% increase variably correlates with P_tcO2 changes [8, 11]. First, we aimed to better study two alternative outcomes (i.e. interrupter resistance (R_int) and S_pO2) and challenge the current recommendations [1] of measuring resistance during inspiration (as opposed to measuring during expiration for reversibility testing [12]), because the physiological expiratory glottis closure can be enhanced during bronchial challenge-induced bronchoconstriction and specific extrathoracic airway reactivity to bronchoconstrictor agents can occur. Second, we wished to evaluate the proposed thresholds for R_int and S_pO2 (+35% and −5% baseline, respectively), as only a 3% decrease is considered to be significant in sleep studies and a mean±sd S_pO2 decrease of −5.2±3.1% corresponds to a much larger than 20% decrease in FEV₁ in 5–8-year-old asthmatic children (−33.3±7.4% decrease in FEV₁) [13].

Between June 2013 and September 2014, we prospectively and consecutively included 28 children unable to correctly perform a spirometry who were referred to our lung function laboratory for a methacholine challenge. Children had to be free of treatment and acute respiratory symptoms for 3 weeks. Chest auscultation had to be normal.

At each step of the bronchial challenge, inspiratory and expiratory series of at least five correct interruptions (R_intinsp and R_intexp, respectively) were performed in random order (but always in the same order with each specific child) using a MicroRint device (Micro Medical, Rochester, UK). P_tcO2 and S_pO2 were recorded throughout the test as previously described [8] using a Tina CombiM (Radiometer, Bronshoj, Denmark). Lung function was checked to be within the range of normal at baseline and assessed after inhalation of saline (diluent) to obtain the reference for changes during the challenge. Doubling doses of methacholine were inhaled, using the dosimeter method, every 5 min [8], from 50 µg up to a cumulative dose of 800 µg. The test ended when P_tcO2 had fallen by 20% or more (PD₂₀P_tcO2), the child had respiratory symptoms or the maximal dose of methacholine was reached. The study was approved by the Institutional Review Board of the French learned society for respiratory medicine (Société de Pneumologie de Langue Française) (CEPRO 2013-015) and the children's parents gave informed consent to the study.

Repeated measurements in children were compared using paired the Wilcoxon signed-rank test. Comparisons of lung function indices between groups of children (responsive and nonresponsive) were performed using the Fisher exact test.

27 (13 girls and 14 boys, median (range) age 5.5 (4.2–8.1) years) children completed all measurements during the bronchial challenge. One child pulled off the P_tcO2 electrode before the end of the test and was, therefore, excluded. 25 children were referred for chronic cough (started at a median age of 2.7 (0.3–8) years), one for suspicion of wheezing and one for dyspnoea upon exertion.

At baseline, R_intexp was higher than R_intinsp (mean 0.81 versus 0.60 kPa·s·L⁻¹, with a mean difference of −0.21 kPa·s·L⁻¹ (95% CI −0.26– −0.16 kPa·s·L⁻¹); p<0.0001), but within the range of normal for all children [14]. At the time of interruption, expiratory airflow was lower than inspiratory airflow throughout the test (e.g. at baseline: 0.30 and 0.39 L·s⁻¹, respectively; p<0.002). 20 children reached the PD₂₀P_tcO2 at a median cumulative dose of methacholine of 100 µg (50–400 µg) (responsive children) without any respiratory symptoms. 14 responsive children had an at least 35% R_intinsp increase (PD₃₅R_intinsp) during the methacholine challenge whereas six responsive children and all the nonresponsive children did not reach PD₃₅R_intinsp (p<0.002). Using R_intexp, there was no association between PD₃₅R_intexp at any time during the test and the presence of BHR (p=1). Therefore, sensitivity and specificity were 70% (95% CI 48–85%) and 100% (95% CI 65–100%), respectively, for R_intinsp, and 50% (95% CI 30–70%) and 57% (95% CI 25–84%), respectively, for R_intexp to detect BHR at or before PD₂₀P_tcO2. Taking into account all cases of discordance between R_int and P_tcO2 changes (significance of the changes at each test step), the number of discordant R_intexp values (n=19) was higher than that of R_intinsp values (n=11) (table 1). For both R_int measurements, the discordances with P_tcO2 changes were equally due to PD₃₅R_int reached before PD₂₀P_tcO2 or to a less than 35% R_int increase at PD₂₀P_tcO2. In the majority of cases, R_intinsp steadily increased during the bronchial challenge, whereas R_intexp had a more irregular pattern of changes and the final change in R_intexp was smaller than that of R_intinsp in all the study children (table 1). All the children (n=11) whose R_intinsp increased by 35% or more without a concomitant 20% P_tcO2 decrease were eventually responsive, whereas three of the nine children with early PD₃₅R_intexp remained nonresponsive throughout the test (three R_intexp false positives). Finally, at PD₂₀P_tcO2, R_intinsp and R_intexp would not have diagnosed BHR in six cases and 10 cases, respectively (false negative), representing 12 children, among whom only two had a 5% decrease in S_pO2 at the same time.

Using R_intinsp changes expressed as percentage of predicted rather than percentage of baseline would have changed the significance of a R_intinsp increase in two out of 81 R_intinsp measurements performed after methacholine inhalation in all study children. These two measurements occurred after the first dose of methacholine in two discordant children (PD₃₅R_intinsp reached before PD₂₀P_tcO2) in whom, after the second methacholine inhalation, both changes (% predicted and % baseline) corresponded but remained discordant with that of PD₂₀P_tcO2. Therefore, the analysis of the concordance between R_intinsp and PD₂₀P_tcO2 changes would not change using percentage predicted or percentage baseline.

If the threshold for R_int were increased by up to 40%, discordance between P_tcO2 and R_intinsp would remain the same, whereas discordance with R_intexp would decrease from 19 to 15 cases (still with two false positives). If a 3% decrease in S_pO2 were the threshold, 15 out of the 20 responsive children would have reached this threshold at PD₂₀P_tcO2 (none before PD₂₀P_tcO2), while none of the nonresponsive children would have reached it at any step of the test (p<0.001). Moreover, using PD₃₅R_intinsp or a 3% decrease in S_pO2 as a composite criterion for bronchial responsiveness, only one responsive child would not have been diagnosed as responsive at PD₂₀P_tcO2 (sensitivity 95%, 95% CI 76–99%) versus six false negatives with PD₃₅R_intinsp or −5% S_pO2 criterion.

Our results do not support a universal physiological mechanism to explain discrepancies between R_int and P_tcO2 measurements during bronchial challenge in young children. The lack of R_int increase in responsive children could reflect an early ventilation–perfusion mismatch with no central airway obstruction but the better concordance between P_tcO2 and R_intinsp over R_intexp remains unexplained. The early reactivity in R_int (before PD₂₀P_tcO2) might be due to glottis changes but we failed to demonstrate any specific recurring patterns of changes of airflow at interruption or of difference between R_intinsp and R_intexp explaining the discrepancies recorded.

To challenge the proposed threshold for R_int [1], we switched from a 35% to a 40% increase and the total number of discordances decreased only for R_intexp although they remained higher than that of R_intinsp. However, as a R_int device may measure only R_intexp, the threshold of 40% may be useful to implement. In children with no R_int increase at PD₂₀P_tcO2, a 3% decrease in S_pO2 better detected BHR than a 5% decrease. The better accuracy of a −3% S_pO2 threshold, over a −5% threshold, increases the safety of associating R_int and S_pO2 measurements when P_tcO2 is not available.

In conclusion, R_intinsp better detects BHR than R_intexp and might better match PD₂₀P_tcO2 changes. Until larger studies confirm these first results, it is reasonable to stick to the proposal of favouring measurement of R_intinsp rather than R_intexp during methacholine challenge. Our findings strengthen the recommendation to associate bronchial reactivity outcomes when P_tcO2 measurement is not available. Finally, the combination of a 35% R_intinsp increase or a 3% S_pO2 decrease might be a useful criterion for detecting BHR with respect to P_tcO2 changes.