Cognitive activity significantly affects the dynamic cerebral autoregulation, but not the dynamic vasoreactivity, in healthy adults

Abstract

Introduction: Neurovascular coupling (NVC) is an important mechanism for the regulation of cerebral perfusion during intensive cognitive activity. Thus, it should be examined in terms of its effects on the regulation dynamics of cerebral perfusion and its possible alterations during cognitive impairment. The dynamic dependence of continuous changes in cerebral blood velocity (CBv), which can be measured noninvasively using transcranial Doppler upon fluctuations in arterial blood pressure (ABP) and CO₂ tension, using end-tidal CO₂ (EtCO₂) as a proxy, can be quantified via data-based dynamic modeling to yield insights into two key regulatory mechanisms: the dynamic cerebral autoregulation (dCA) and dynamic vasomotor reactivity (DVR), respectively.

Methods: Using the Laguerre Expansion Technique (LET), this study extracted such models from data in supine resting vs cognitively active conditions (during attention, fluency, and memory tasks from the Addenbrooke's Cognitive Examination III, ACE-III) to elucidate possible changes in dCA and DVR due to cognitive stimulation of NVC. Healthy volunteers (n = 39) were recruited at the University of Leicester and continuous measurements of CBv, ABP, and EtCO₂ were recorded.

Results: Modeling analysis of the dynamic ABP-to-CBv and CO₂-to-CBv relationships showed significant changes in dCA, but not DVR, under cognitively active conditions compared to resting state.

Discussion: Interpretation of these changes through Principal Dynamic Mode (PDM) analysis is discussed in terms of possible associations between stronger NVC stimulation during cognitive tasks and enhanced sympathetic activation.

Keywords: cerebral blood velocity; dynamic cerebral autoregulation; dynamic vasomotor reactivity; neurovascular coupling; transcranial doppler ultrasonography.

FIGURE 1

Block diagram of the employed linear time-invariant model with ABP and EtCO₂ inputs (extracted from the raw signals) convolving the respective kernels to generate the model-predicted CBv output as the sum of the two convolutions (denoted by the Ⓧ symbol). The ABP and EtCO₂ kernels describe the linear dynamics of each input-output pathway and are estimated via LET.

FIGURE 2

The obtained average kernel estimates (A, C) and model-predicted step responses (B, D) [in cm/(mmHg*s²)] for the ABP-to-CBv (A, B) and CO₂-to-CBv (C, D) relations under baseline and active conditions with standard deviation bounds (dashed lines). (C) Notable changes are seen only in the size of the CO₂ kernel that is reduced significantly during cognitively active conditions (COG), while retaining the basic waveform.

FIGURE 3

Average Gain Functions (A, C) [in cm/(mmHg*s)] and average Phase Functions (B, D) [in radians] for the ABP-to-CBv kernels (A, B) and CO₂-to-CBv kernels (C, D) under resting (BL) and active (COG) conditions with standard deviation bounds (dashed lines). (A) The average ABP Gain Function exhibits lower values below 0.04 Hz and a small bulge around 0.07 Hz for BL conditions. Both average Gain Functions exhibit a primary resonant peak around 0.2 Hz and have similar magnitude. (C) The average CO₂ Gain Functions exhibit similar low-pass characteristics, but slightly reduced under COG condition. The maximum value of the ABP Phase Function (B) is reduced considerably for the COG condition.

FIGURE 4

PDMs for the ABP-to-CBv relation (A, B) and CO₂-to-CBv relation (C, D) in the time-domain (A, C) and their frequency-domain counterparts, magnitude only (B, D) [in cm*sec^-1/mmHg].