The major limitation to exercise performance in COPD is inadequate energy supply to the respiratory and locomotor muscles.

No doubt dynamic hyperinflation and lack of oxidative capacity of skeletal muscles are important causes of exercise limitation in COPD. O’Donnell and Webb and Debigaré and Maltais will convince the reader of this by the elegant experiments they have performed. The thesis we will put forward is that during the natural history of COPD the primary factors leading to impairment of exercise performance are an increase in energy demands combined with a decrease in supplies and that both of these result from excessive recruitment of expiratory muscles. We argue that both dynamic hyperinflation and reduced oxidative capacity are secondary adaptations resulting from this primary abnormality. Increased energy demands during exercise in COPD. Energy demands are increased in COPD because of the high O2 cost of breathing (V̇O2resp). In health, V̇O2resp is only 1–3 ml O2/l breathed, whereas in COPD it has been reported variously to average 6.3, 9.7, and 16.4 ml/l breathed with individual values ranging from 3.0 to 19.5 ml/l (8, 17). The large between patient range in V̇O2resp probably reflects variation in the work of breathing (Wresp). During exercise, a large variation in Wresp certainly exists. In two studies, COPD patients formed two distinct groups: those that strongly recruited abdominal muscles and those that did not (5, 10). In the first (5), at an exercise workload of 10 W the work performed on the lung averaged 754 cmH2O l 1 min 1 in recruiters but only 277 cmH2O l 1 min 1 in nonrecruiters, although ventilation was similar. Expiratory muscle activation is the normal response to exercise (1) so the recruiters behaved normally. The problem is that in COPD, it fails to increase ventilation, because expiratory flow becomes limited by high pleural pressures. While abdominal muscle recruitment is beneficial during exercise in health (1), it is definitely harmful in COPD (4, 5). Because Wresp was 2.7-fold greater in recruiters, we can assume that their V̇O2resp was twice as high as the nonrecruiters. Let’s also assume that it was 12 ml O2/l in the former and 6 ml/l in the latter. The maximal exercise workload (Wmax) was 20 and 35 W in recruiters and nonrecruiters (P 0.05), while V̇E at Wmax was 35.9 and 37.9 l/min, respectively (5). Thus the estimated V̇O2resp was 430.8 ml/min in recruiters but only 227.4 ml/min in nonrecruiters. From the measured values of V̇O2 at rest and during 10 W exercise and assuming that V̇O2 increased linearly (dV̇O2/dwatt is constant) the V̇O2 at maximal exercise workload (V̇O2max) was 830.0 and 1,327.5 ml O2/min, respectively, in recruiters and nonrecruiters. Subtracting V̇O2resp from V̇O2max reveals that if the respiratory muscles received all their demands there was only 399.2 ml O2 available to locomotor muscles and other body tissues in recruiters but 1,100.1 ml in nonrecruiters. The respiratory muscles demanded 53% of V̇O2max in recruiters but only 17%, a value close to normal (6), in nonrecruiters. The nonrecruiters’ breathing pattern was abnormal because abdominal muscles were not recruited during exercise. As a result, their exercise performance was better. However, their resting lung function was worse. Both the FEV1 and FEV1/ FVC were significantly lower in nonrecruiters. This strongly suggests that as COPD progresses, patients eventually realize that abdominal muscles recruitment is bad and somehow they learn to derecruit them. Alas, without abdominal muscle contraction they dynamically hyperinflate. They can exercise a bit more, but not much (15). Thus we believe that dynamic hyperinflation results from a learned response to an inadequate supply of energy to meet demands. Decreased energy supplies during exercise with expiratory flow limitation. When normal subjects breathe with a Starling resistor in the expiratory line, which limits expiratory flow to 1 l/s, exercise is limited by severe dyspnea; abdominal pressure (Pab) increases abnormally; duty cycle decreases; CO2 retention occurs, increasing Pab even more (3, 13, 14); the high expiratory pressures and short duty cycle act like a Valsalva maneuver and decrease cardiac output (Q c) (2); as a result, O2 debt is increased by 52% (22). Expiratory flow limitation (EFL) decreases the shortening velocity of abdominal muscles, and, in accordance with their force velocity characteristics Pab increases (3). Expiratory muscle recruitment can account for 66% of the variation in Borg scale ratings of difficulty in breathing (14). None of these abnor