Reliability and agreement between alternative CO2 paradigms for assessment of cerebrovascular reactivity in cerebral small vessel disease.

Cerebrovascular reactivity (CVR) to CO2 is increasingly used as an index of cerebrovascular health and function in trials in cerebral small vessel disease (CSVD). Further work is needed to determine the validity and optimal parameters of this technique. In 75 CSVD patients in the placebo-controlled, crossover OxHARP trial of sildenafil, agreement and reproducibility of baseline (1 st visit) and placebo visit CVR was derived comparing 4% and 6% CO2 exposure. Peak response to each dose was used to estimate CVR (Δ cerebral blood velocity/ Δ endtidal CO₂) using absolute (cm.s -1 /etCO2 %), fully standardised (%Δ/%Δ) or semi-standardised (%Δ/etCO2 %) approaches, and compared to CVR from delay-shifted, beat-to-beat associations from general linear models (GLM). Single-point measurements showed less inter-and intra-participant variability. The 6% CO2 prompted higher absolute and standardised CVR values, although this was not observed with the semistandardised of analysis. CVR estimates using absolute values showed comparable inter-and intraparticipant variability across CO₂ challenges. However, reproducibility improved when CVR values were standardised following exposure to 6% CO2 or derived using GLM that incorporated data from both 4% and 6% CO2 data. The strongest agreement between CVR methods was achieved with analyses using absolute values. CVR assessments were reproducible, but this reliability was influenced by CO₂ dose and the analytical approach. While single-point measurements showed the best inter-and intra-participant variability, overall, CVR estimation using a GLM demonstrated a superior test-retest reliability across visits.