Individual Patient Data Meta-Analysis of Dynamic Cerebral Autoregulation and Functional Outcome After Ischemic Stroke

Abstract

Background: The relationship between dynamic cerebral autoregulation (dCA) and functional outcome after acute ischemic stroke (AIS) is unclear. Previous studies are limited by small sample sizes and heterogeneity.

Methods: We performed a 1-stage individual patient data meta-analysis to investigate associations between dCA and functional outcome after AIS. Participating centers were identified through a systematic search of the literature and direct invitation. We included centers with dCA data within 1 year of AIS in adults aged over 18 years, excluding intracerebral or subarachnoid hemorrhage. Data were obtained on phase, gain, coherence, and autoregulation index derived from transfer function analysis at low-frequency and very low-frequency bands. Cerebral blood velocity, arterial pressure, end-tidal carbon dioxide, heart rate, stroke severity and sub-type, and comorbidities were collected where available. Data were grouped into 4 time points after AIS: <24 hours, 24 to 72 hours, 4 to 7 days, and >3 months. The modified Rankin Scale assessed functional outcome at 3 months. Modified Rankin Scale was analyzed as both dichotomized (0 to 2 versus 3 to 6) and ordinal (modified Rankin Scale scores, 0-6) outcomes. Univariable and multivariable analyses were conducted to identify significant relationships between dCA parameters, comorbidities, and outcomes, for each time point using generalized linear (dichotomized outcome), or cumulative link (ordinal outcome) mixed models. The participating center was modeled as a random intercept to generate odds ratios with 95% CIs.

Results: The sample included 384 individuals (35% women) from 7 centers, aged 66.3±13.7 years, with predominantly nonlacunar stroke (n=348, 69%). In the affected hemisphere, higher phase at very low-frequency predicted better outcome (dichotomized modified Rankin Scale) at <24 (crude odds ratios, 2.17 [95% CI, 1.47-3.19]; P<0.001) hours, 24-72 (crude odds ratios, 1.95 [95% CI, 1.21-3.13]; P=0.006) hours, and phase at low-frequency predicted outcome at 3 (crude odds ratios, 3.03 [95% CI, 1.10-8.33]; P=0.032) months. These results remained after covariate adjustment.

Conclusions: Greater transfer function analysis-derived phase was associated with improved functional outcome at 3 months after AIS. dCA parameters in the early phase of AIS may help to predict functional outcome.

Keywords: acute stroke; autoregulation; cerebrovascular stroke; meta-analysis; transcranial Doppler ultrasonography.

Graphical abstract

Figure 1

Dynamic Cerebral Autoregulation (dCA) determined via transfer function analysis (TFA). Panel A illustrates the conversion of cerebral blood velocity (CBv) and mean arterial pressure (MAP) from the time to frequency domain to derive metrics of coherence, gain and phase. Panel B illustrates the phase shift (A) between the blood pressure and CBv waveforms, and the difference in amplitude (gain, B) between the blood pressure and CBv waveforms. MCAv=middle cerebral artery blood velocity. Figure adapted from^, with permission.

Figure 2

Mean phase at very low frequency (VLF) (A), autoregulation index (ARI) (C) in the affected hemisphere (AH) within 24h in participants with good (modified Rankin Scale [mRS]:0–2) vs poor [mRS 3–6]) outcome at 3mo. The predicted probability of good vs poor outcome with increasing phase at VLF (B), and ARI (D).

Figure 3

Mean phase at very low frequency (VLF) (A), autoregulation index (ARI) (C) in the affected hemisphere (AH) at 24–72h in participants with good (modified Rankin Scale [mRS]:0–2) vs poor (mRS:3–6) outcome at 3mo. The predicted probability of good vs poor outcome with increasing phase at VLF (B), and ARI (D).