The Pressure Reactivity Index as a Measure for Cerebrovascular Autoregulation: Validation in a Porcine Cranial Window Model.

ABSTRACT

BACKGROUND AND OBJECTIVES: Pressure reactivity index (PRx) has been proposed as a metric associated with cerebrovascular autoregulatory (CA) function and has been thoroughly investigated in clinical research. In this study, PRx is validated in a porcine cranial window model, developed to visualize pial arteriolar autoregulation and its limits. METHODS: We measured arterial blood pressure, intracranial pressure, pial arteriolar diameter, and red blood cell (RBC) velocity in a closed cranial window piglet model during gradual balloon catheter-induced arterial hypotension (n = 10) or hypertension (n = 10). CA limits were derived through piecewise linear regression of calculated RBC flux vs cerebral perfusion pressure (CPP), leading for each arteriole to 1 lower limit of autoregulation (LLA) and 2 upper limits of autoregulation (ULA1 and ULA2). Autoregulation limits were compared with PRx thresholds, and receiver operating curve analysis was performed with and without CPP binning. A linear mixed effects model of PRx was performed. RESULTS: Receiver operating curve analysis indicated an area under the curve (AUC) for LLA prediction by a PRx of 0.65 (95% CI 0.64-0.67) and 0.77 (95% CI 0.69-0.86) without and with CPP binning, respectively. The AUC for ULA1 prediction by PRx was 0.69 (95% CI 0.68-0.69) without and 0.75 (95% CI 0.68-0.82) with binning. The AUC for ULA2 prediction was 0.55 (95% CI 0.55-0.58) without and 0.63 (95% CI 0.53-0.72) with binning. The sensitivity and specificity of binned PRx were 65%/90% for LLA, 69%/71% for ULA1, and 59%/74% for ULA2, showing wide interindividual variability. In the linear mixed effects model, pial arteriolar diameter changes were significantly associated with PRx changes (P = .002), whereas RBC velocity (P = .28) and RBC flux (P = .24) were not. CONCLUSION: We conclude that PRx is predominantly determined by pial arteriolar diameter changes and moderately predicts CA limits. Performance to detect the CA limits varied highly on an individual level. Active therapeutic strategies based on PRx and the associated correlation metrics should incorporate these limitations.