Indexes of cerebral autoregulation do not reflect impairment in syncope: insights from head-up tilt test of vasovagal and autonomic failure subjects

Abstract

Purpose

The study of dynamic cerebral autoregulation (CA), which adapts cerebral blood flow to arterial blood pressure (ABP) fluctuations, has been limited in orthostatic intolerance syndromes, mainly due to its stationary prerequisites hardly to meet during maneuvers to provoke syncope itself. New techniques of continuous estimates of CA could overcome this pitfall. We aimed to evaluate CA during head-up tilt test in common conditions causing syncope.

Methods

We compared three groups: eight controls; eight patients with autonomic failure due to familial amyloidotic polyneuropathy; eight patients with vasovagal syncope (VVS). ABP and cerebral blood flow velocity (CBFV) were measured with Finometer^® and transcranial Doppler. We calculated cerebrovascular resistance index (CVRi), critical closing pressure (CrCP) and resistance area product (RAP), and derived CA continuously from autoregulation index [ARI(t)].

Results

With HUTT, AF subjects showed a pronounced decrease in CBFV (−36 ± 17 versus −7 ± 6%, p < 0.0001), ABP (−29 ± 27 versus 7 ± 12%, p < 0.0001) and RAP (−17 ± 23 versus 3 ± 18%, p < 0.0001) but not CVRi (p = 0.110). VVS subjects showed progressive cerebral vasoconstriction prior to syncope, (reduced CBFV 19 ± 15 versus 1 ± 6, p < 0.000; increased RAP 12 ± 18 versus 2 ± 3%, p = 0.024 and CVRi 12 ± 18 versus 2 ± 3%, p = 0.005). ARI(t) increased significantly in AF patients (5.7 ± 1.2 versus 6.9 ± 1.2, p = 0.040) and VVS (5.8 ± 1.2 versus 7.3 ± 1.2, p = 0.015) in response to ABP fall during syncope.

Conclusions

Our data suggest that dynamic cerebral autoregulatory response to orthostatic challenge is neither affected by autonomic dysfunction nor in neutrally mediated syncope. This study also emphasizes that RAP + CrCP model is more informative than CVRi, mainly during cerebral vasodilatory response to orthostatic hypotension.

Appendix

Critical closing pressure and resistance area product calculations

After the first harmonic is derived from ABP (P1) and CBFV (V1) for each cardiac cycle, the resistance area product (RAP) can be calculated as:

$${\text{RAP}} = \frac{{{\text{P}}1}}{{{\text{V}}1}}.$$

(1)

And the critical closing pressure (CrCP) can then be obtained as:

$${\text{CrCP}} = {\text{MAP}} - V_{\text{mean}} \cdot {\text{RAP,}}$$

(2)

where MAP is the mean ABP for the cardiac cycle and V _mean the corresponding mean CBFV.

Transfer function analysis

From beat-to-beat values of mean CBFV and mean ABP, we can obtain the corresponding auto-spectra, $S_{xx}$ and $S_{yy}$, with the fast Fourier transform, as well as the cross-spectra, $S_{xy}$, with the Welch method (Claassen et al. 2016). We deduce the oscillatory influence of ABP over CBFV at a given frequency $f$ by the transfer function $H$:

$$H(f) = \frac{{S_{xy} (f)}}{{S_{xx} (f)}}.$$

(3)

From its real $H_{\text{R}}$ and imaginary parts $H_{\text{I}}$ we can then derive:

$${\text{gain}},\quad \;\left| {H(f)} \right| = \sqrt {(\left| {H_{\text{R}} (f)} \right|^{2} + \left| {H_{\text{I}} (f)} \right|^{2} )} ,$$

(4)

$${\text{phase}},\;\quad \varphi = \tan^{ - 1} \left( {\frac{{H_{\text{I}} (f)}}{{H_{\text{R}} (f)}}} \right).$$

(5)

T To assess how much of the output (CBFV) power is explained by the corresponding input (ABP) power at each frequency, we calculate the coherence between spectra $(\gamma^{2} (f))$ by the formula:

$$\gamma^{2} = \frac{{\left| {S_{xy} (f)} \right|^{2} }}{{\left| {S_{xy} (f) \cdot S_{yy} (f)} \right|}}.$$

(6)

ARMA implementation of Tiecks’ model

The model proposed by Tiecks et al. (1995) uses a second-order differential equation to predict the velocity signal V(t) corresponding to a relative pressure change given by dP(t) by the formula:

$$V(t) = 1 + {\text{d}}P(t) - K \times x_{2} (t),$$

(7)

where K represents a gain parameter in the second-order differential equation, an x ₂(t) is a state variable obtained from the following state equation system representing a second-order linear differential equation modeled by gain (K), time constant (T) and dampening factor (D). So, for a gives sampling frequency f and each sample discrete value n:

$$x2n = x2n - 1 - \frac{x1 - 2D \cdot x2n - 1}{f \cdot T},$$

(8)

$$x1n = x1n - 1 + \frac{{{\text{d}}Pn - x2n - 1}}{f \cdot T}.$$

(9)

In the original proposal of Tiecks et al. (1995), only 10 combinations of the parameters K, D, and T were considered, according to the values given in their Table 3, which also shows the corresponding value of ARI for each combination of these parameters.

The building of an ARMA model based on Tiecks’ model is extensively detailed in previous work (Dineen et al. 2010). For the sake of simplicity, we resume main formulas of the method here. Firstly we express the transfer function as Z-transforms, and then applying an inverse Z transform to derive:

$$v(n) = a \times p(n) + b\left[ {p(n - 1) - v(n - 1)} \right] + c[p(n - 2)v(n - 2)] ,$$

(10)

where p(n) and v(n) are discrete samples of V(t) and P(t), respectively.

Once the ARMA parameters have been estimated, the CBFV step response can be obtained from Eq. (10), and the ARI parameter can then be extracted by least squares fitting of the corresponding Tiecks et al. (1995) model responses using the first N _fit samples of the ARMA step response.