Lactates kinetics: From the serial measurements of lactate concentrations during exercise, the La/VO2 relationship was analyzed for each subject as previously described . Briefly, an exponential function,
La = a + bV°2
where a + 1 is the intercept on the y axis and b is a dimension-less parameter describing the steepness of the relationship, was used to fit the experimental data. The curve fitting was constrained so that parameter a could not be smaller than -1 to avoid negative values for La. This model was used because it provided a parameter b from which the steepness of the La/VO2 relationship could be readily compared among individuals. In a previous study the authors found that the La/VO2 relationship was well fitted by this simple model in patients with COPD . Functional status classification: The functional status was graded according to the VO2 peak using the approach previously used in chronic heart failure by Weber and Janicki : greater than 16 mL/min/kg and 20 mL/min/kg or less indicated mild to moderate impairment (class B, n=15), greater than 10 mL/min/kg and 16 or less, moderate to severe impairment (class C, n=31), and greater than 6 mL/min/kg and 10 or less, severe impairment (class D, n=8).
Statistical analysis: Values are reported as mean ± SD unless otherwise stated. The maximal voluntary ventilation (MVV) was estimated by multiplying the forced expiratory volume in 1 s (FEV1) by 35 . Comparisons between groups were done using ANOVA and Tukey’s comparison test. Parameters obtained from the curve fitting of the La/VO2 relationships were estimated using an iterative method (Marquardt-Levenberg algorithm). Regression analyses were done using the least squares method. P<0.05 was considered statistically significant.