Middle cerebral artery dynamic cerebral autoregulation is impaired by infarctions in the anterior but not the posterior cerebral artery territory in patients with mild strokes

1 Introduction

Cerebral perfusion depends on various factors, such as the mean arterial pressure (MAP), cranial perfusion pressure, blood carbon dioxide and oxygen tension, blood pH value, metabolic demands, and remote influences from the brain’s functional connectivity networks (CNs) [1]. In a strict sense, the relationship between cerebral blood flow (CBF) and MAP determines cerebral autoregulation (CA); the other variables modulate this relationship, influencing overall CBF. The phenomenon of neurovascular coupling (NVC) is illustrated by cognitive tasks and arm movements that increase CBF or its velocity (CBFV, measured by transcranial Doppler [TCD] ultrasound), indicating increased perfusion [2,3,4,5,6,7]. Interestingly, such NVC studies showed additional effects in various other vascular territories, ultimately leading to the concept of brain CNs [2,3,6,7].

During an ischaemic stroke in the middle cerebral artery (MCA) territory, the dynamical aspects of CA (dCA) are frequently disturbed; interestingly, dCA is disturbed not only on the stroke-affected side but also on the unaffected side [8]. The reason for this behaviour is unclear, but systemic factors such as high blood pressure (BP) [9,10], diabetes mellitus [11], or the presence of cerebral microangiopathy [12,13] are discussed. CN considerations as a possible confounder of the dCA in the MCA have not yet been addressed. To gain a first impression on such a hypothesis, we investigated whether isolated infarctions in other supratentorial vascular territories (anterior cerebral artery [ACA]; posterior cerebral artery [PCA]) can affect dCA in the MCA. Given the body of knowledge about the motoric CNs derived from functional magnetic resonance imaging (fMRI; summarized in refs [2,3]), we hypothesize that infarctions in the motoric areas in the ACA territory should affect the dCA in the MCA, while infarctions in the PCA territory which usually do not have motor symptoms would not. Anatomically, the blood supply to the PCA is usually provided via the vertebral and the basilar arteries, and both are independent of the internal carotid artery (ICA) and the MCA. The ICA supplies both the MCA and the ACA. To avoid any contamination of our results by haemodynamically relevant ICA obstructions, we considered this possibility a priori and excluded patients with such relevant ICA pathologies.

2 Methods

The Lucerne Hospital is a large tertiary teaching hospital with a full-service stroke centre. All patients with stroke syndrome receive standardized care, first undergoing a focused clinical examination followed by multimodal cranial computed tomography. If indicated, intravenous thrombolysis and/or arterial thrombectomy immediately follows. All patients with stroke syndrome are transferred to the stroke unit for close clinical monitoring. Patients are scored on the National Institutes of Health Stroke Scale (NIHSS [14]) and modified Rankin scale [15] upon hospital admission and daily while in the stroke unit; they also undergo monitoring of MAP, heart rhythm, body temperature, blood glucose concentration, and oxygen saturation. An echocardiogram measuring the left ventricular ejection fraction, an extensive ultrasound examination with assessment of dCA, brain MRI with diffusion-weighted imaging (DWI), T2, fluid-attenuated inversion recovery, and susceptibility-weighted imaging sequences were performed within a maximum of 72 h after hospitalization. Summarizing all imaging modalities, each ischaemic event is classified according to the Trial of Org 10172 in Acute Stroke Treatment (TOAST) classification [16]. If the neurological deficit resolved within 24 h and DWI remained negative, these patients were classified as having suffered a transient ischaemic attack. Irrespective of whether the patient is symptomatic, any present cerebral microangiopathy is classified according to the Fazekas scale [17]. Infarct size is calculated by the ABC/2 method, which we evaluate manually and compare with the measurements made by automated software [18].

For this study, we recruited 10 consecutive patients with unilateral ACA infarction and 22 with PCA infarction who had received the abovementioned structured stroke care, who had temporal bone windows for TCD ultrasound examinations, and who did not have stenosis ≥50% in the extra- or intracranial arteries supplying the brain. All investigations were performed within 72 h after symptom onset. From our previously reported healthy population [19], we included 29 age- and sex-matched participants as controls and averaged the results of the bilateral MCAs as a reference value. All healthy persons were without known cardiocirculatory complains or anatomical variations.

3 dCA assessment

For details of the dCA assessment, see previous reports [3,19]. CBFV (MultidopX, DWL; Compumedics, Sipplingen, Germany; 2-MHz probes) and BP (Finometer Pro; Finapres Medical Systems, Amsterdam, The Netherlands) were simultaneously recorded over a minimum period of 6 min. The end-tidal pCO₂ (ETCO₂) was measured via nostril tubes using a capnography function built into the TCD device. The ETCO₂ for each patient is reported as the mean ETCO₂ over the total recording period. Cerebrovascular resistance was calculated by the mean BP over the mean CBFV. Patient recordings were performed at their stroke unit site with the patients in a resting position and the head slightly elevated. They were instructed not to make movements. Those of the controls (healthy subjects) were performed in a separate room with the same resting body position and the head slightly elevated. All recordings followed the recommendations of the CA network as much as possible [20].

The data were analysed using MATLAB (2020b; MathWorks Inc., Natick, MA, USA). After visual inspection for artefacts, only artefact-free data periods of 5 min were used. Each raw data time series was averaged over 1-s intervals (one example of a patient with an ACA infarction is given in Figure 1). The transfer function estimates between BP and CBFV time series were extracted from their respective power autospectra or cross-spectra using Welch’s averaged periodogram method, with a Hanning window length of 100 s, a window overlap of 50%, and a total fast Fourier transformation data length of 300 s. Coherence, phase (in radians), and gain (in cm/s/mmHg or as %gain in %cm/s/mmHg) were calculated over a frequency range of 0.02–0.50 Hz and are reported as their respective average in the very low (VLF, 0.02–0.07 Hz), low (LF, 0.07–0.15 Hz), and high (HF, 0.16–0.5 Hz) frequency ranges. For interpretation, gain and phase usually follow an opposite direction, and an impaired dCA is indicated by a low phase and/or a high gain.

Figure 1

A representative recording of BP and MCA blood flow velocity in both hemispheres of a patient with an infarction in the ACA. All time series are averaged over 1 s to reduce the amount of data points. The calculated phase values were as follows: in the very low frequency 0.54 radian in the affected hemisphere and 0.64 in the unaffected one; in the low frequency phase 0.43 in both MCAs.

4 Results

The overall baseline characteristics of the patients and the controls are reported in Table 1. None of the presented baseline characteristics significantly predicted the dCA parameter by regression analysis or showed significant differences between the patients with PCA or ACA infarcts. Only infarct size (see below) was different between the PCA and ACA infarctions (p < 0.01). The allocations of stroke-related large artery disease were in the vertebral artery system as we excluded patients with large artery disease in the carotid artery system.

In the patients with PCA infarctions, the median infarct volume was 4.10 ml [0.40, 12.75], with a maximum of 106 ml. Motor symptom was not present in any patient with PCA infarct. Considering the dCA parameters, there were no significant differences between the PCA infarct patients and the controls (Table 2). Thus, infarct size seems to be unrelated to dCA in the MCA.

In the patients with ACA infarctions, the median infarct size was 0.9 ml [0.04, 4.08], with a maximum of 29 ml. Regression analysis showed that dCA parameters were not related to infarct size. Compared to the controls (Figure 2), phase was significantly reduced in the VLF and the LF in the affected and unaffected hemispheres; the result that phase in the nonaffected side was lower than that in the affected side was mainly due to a very low outliner; phase was similarly low on both sides when this outliner was removed; therefore, our result regarding the unaffected side should not be overinterpreted given the small total number of patients. There was little difference in gain except for an increased gain in the unaffected hemisphere in the VLF range (due to the same outliner patients as in phase). Figure 3 demonstrates that nearly all infarct lesions were within the motor or premotor cortical areas. Clinically, all patients exhibited hemiparesis with the leg weaker than the arm (seven due to left-sided lesions and three due to right-sided lesions).

Figure 2

Comparison of MCA phase between controls and the patients with infarcts in the PCA or the ACA territories in the very low (VLF) and low frequency (LF) ranges.

Figure 3

Location (n = 10) and horizontal extent of the infarcts in the ACA territory. Each colour represents one patient; the infarctions were allocated in the right hemisphere in three patients and in the left hemisphere in seven patients. In one patient the coloured dashed line was chosen to indicate infarct size to allow us to visualize the other infarcts which are within the boundaries of this large infarct.

5 Discussion

Overall, PCA infarction does not affect CBF regulation in the MCA territory, while ACA infarction does with pronounced leg hemiparesis as the major motor finding. Because the main cardiovascular factors and ETCO₂ were not different between the two stroke allocation groups, they cannot account for the different dCA findings. The sizes of infarcts in both territories were also unrelated to the dCA impairments in the MCA territory; large artery disease in the carotid artery distribution (as a common reason of dCA impairment) was excluded, and small vessel disease was rare. As Figure 1 shows, all our ACA lesions affected cortical areas relevant for motor function. We, therefore, interpret our findings as supportive for our hypothesis that motor CN contribute to MCA dCA disturbances [4,5,21,22,23].

To our knowledge, our study is the first one that describes with a bed side test that cerebral perfusion regulation in the MCA territory is affected in patients with ACA infarctions, while CBFV, as an index of CBF, is unchanged. Thus, hypothetically, dCA assessment seems more sensitive for studying even fine perfusion regulation processes and could be used for a wider approach to study motor recovery predictability [23]. Compared to MRI or other CBF-measuring techniques, the advantages of assessing dCA by TCD are its robustness and its availability as a bedside test.

This study has limitations. The environments were different between the patients and the controls. This could have influenced the results as the stroke-unit setting is noisy and could produce more noisy recordings, making the recordings more sensitive to errors [24]. However, the clear distinction between the results in the PCA infarct patients and the ACA infarct patients can be used as an argument against the relevance of the suggested noisy surroundings. The number of patients in the ACA infarct group is small, but the results are very impressive. The mechanism behind the resulting phase decrease is not directly derivable from our data, but the fact that gain as an index of the vascular tone remained widely unchanged favours a metabolic stimulus.