, 2004b). In that study as

in the present one NIV did not influence MEP latency. In the current study in COPD patients, although there was a reduction in diaphragm MEPTS during NIV, there was no significant change in the response to paired stimuli. This suggests that the reduction VX-770 supplier in MEP was principally mediated at a level below the motor cortex. Since isocapnia was maintained this would point to a role for neuromechanical feedback operating either at the spinal level where motor neurons can be preactivated by muscle afferents (Komori et al., 1992) or indirectly via the brainstem respiratory centers which also have afferent input. It has been demonstrated in healthy subjects that inspiratory pressure support ventilation causes hyperventilation since tidal volume rises but respiratory rate does not fall leading to a net fall in CO2 (Lofaso et al., 1992). Interestingly hyperventilation with NIV has not been observed during sleep (Morrell et al., 1993) which implies a role for cortical influences. NIV is associated with a reduction in inspiratory activity assessed

using diaphragm EMG, which persists even if CO2 is corrected (Fauroux et al., 1998), and NIV increases the threshold where a ventilatory response to CO2 occurs (Scheid et al., 1994 and Simon Alpelisib mouse et al., 1991). Using PET measurements of cerebral blood flow it has been shown that a number of cortical areas are involved in the response to increases in inspiratory load ( Isaev et al., 2002) (a response which is itself attenuated by sleep) ( Santiago et al., 1981), however the diaphragm motor cortex itself was not identified although this may have been at a level below the sensitivity of the test used. Because it is not possible to analyze H-reflex or F-waves for the phrenic nerve it is difficult to

assess spinal facilitation directly. The absence of change in intracortical circuits in response to NIV may represent metaplasticity many (Abraham and Bear, 1996), which is a change in the capacity to express plasticity caused by prior exposure; in COPD possibly chronic blood gas derangements or load capacity imbalance in the respiratory muscle pump could be responsible. In the period of spontaneous breathing following NIV, we did not find any change in cortical responses measured compared to baseline. We acknowledge that diaphragm MEP recordings from chest wall electrodes may have been contaminated by signals from either intercostal or abdominal muscles. This was minimized by positioning the surface electrodes close together and optimizing their position in each patient using phrenic nerve stimulation. An alternative would have been to use an esophageal electrode but this would have added significantly to the discomfort of what was already a demanding study for quite severely disabled patients.