High-dose methacholine inhalation tests were carried out using a modification of the method described by Chai et al.17 Methacholine (Sigma Diagnostics; St. Louis, MO) solutions were prepared at different concentrations (0.075, 0.15, 0.3, 0.625, 1.25, 2.5, 5, 10, 25, 50, 100, 150, and 200 mg/mL) in buffered saline solution (pH 7.4). Lung function was measured using a computerized spirometer (Microspiro-HI 298; Chest; Tokyo, Japan), and the largest FEV1 among triplicate values at each time was used for analysis. A Rosenthal-French dosimeter (Laboratory for Applied Immunology; Baltimore, MD), triggered by a solenoid valve set to remain open for 0.6 s, was used to generate an aerosol from a nebulizer (DeVilbiss 646; DeVilbiss Health Care; Somerset, PA), using pressurized air at 20 pounds per square inch. Each subject inhaled five inspiratory capacity breaths of buffered saline solution and increasing concentrations of methacholine at 5-min intervals. This set-up produced an output of 0.009 ± 0.0014 mL (mean ± SD) per inhalation. FEV1 was measured 60 to 90 s after inhalation at each concentration. The procedure was terminated when the FEV1 level fell below 50% of the post-saline solution value, or when a maximal response plateau had been established. This was considered to have occurred if three or more data points at the highest concentrations fell within a 5% response range.,8 An additional 5 or 10 inhalations of the 200 mg/mL solution were taken if the last three data points, showing less than a 50% FEV1 fall, did not satisfy the above criteria. For safety reasons, subjects were given the opportunity to stop the challenge test if they had too much discomfort. Response, expressed as the percentage fall in FEV1 from the post-saline value, was plotted against the log of the inhaled methacholine concentration. The dose-response curves were characterized by their position and maximal response, the former was expressed as PC20, which was calculated by log-linear interpolation between two adjacent data points, and the latter was defined as the level of the maximal response plateau obtained by averaging the consecutive points on the plateau or as the last data point of the dose-response curve if a plateau could not be obtained.