Multivariate dynamic analysis of cerebral blood flow regulation in humans.

The contributions of beat-to-beat changes in mean arterial blood pressure (MABP) and breath-by-breath fluctuations in end-tidal CO2 (EtCO2) as determinants of the spontaneous variability of cerebral blood flow velocity (CBFV) were studied in 16 normal subjects at rest. The two input variables (MABP and EtCO2) had significant cross-correlations with CBFV but not between them. Transfer functions were estimated as the multivariate least mean square finite impulse response causal filters. MABP showed a very significant effect in explaining CBFV variability (p < 10(-11), Fisher's aggregated-p test) and the model mean square error was significantly reduced (p < 0.001) by also including the contribution of EtCO2. The estimated mean CBFV step response to MABP displayed the characteristic return to baseline caused by the cerebral autoregulatory response. The corresponding response to EtCO2 showed a gradual rise taking approximately 10 s to reach a plateau of 2.5%/mmHg. This study demonstrated that spontaneous fluctuations in EtCO2 can help to explain the CBFV variability at rest if appropriate signal processing techniques are employed to address the limited power and bandwith of the breath-by-breath EtCO2 signal.