Regional disparity in continuously measured time-domain cerebrovascular reactivity indices: a scoping review of human literature

Objective: Cerebral blood vessels maintaining relatively constant cerebral blood flow (CBF) over wide range of systemic arterial blood pressure (ABP) is referred to as cerebral autoregulation (CA). Impairments in CA expose the brain to pressure-passive flow states leading to hypoperfusion and hyperperfusion. Cerebrovascular reactivity (CVR) metrics refer to surrogate metrics of pressure-based CA that evaluate the relationship between slow vasogenic fluctuations in cerebral perfusion pressure/ABP and a surrogate for pulsatile CBF/cerebral blood volume. Approach: We performed a systematically conducted scoping review of all available human literature examining the association between continuous CVR between more than one brain region/channel using the same CVR index. Main Results: In all the included 22 articles, only handful of transcranial doppler (TCD) and near-infrared spectroscopy (NIRS) based metrics were calculated for only two brain regions/channels. These metrics found no difference between left and right sides in healthy volunteer, cardiac surgery, and intracranial hemorrhage patient studies. In contrast, significant differences were reported in endarterectomy, and subarachnoid hemorrhage studies, while varying results were found regarding regional disparity in stroke, traumatic brain injury, and multiple population studies. Significance: Further research is required to evaluate regional disparity using NIRS-based indices and to understand if NIRS-based indices provide better regional disparity information than TCD-based indices.


Introduction
Cerebral autoregulation (CA) refers to the cerebral blood vessels' capacity to maintain a relatively constant cerebral blood flow (CBF) over a wide range of systemic arterial blood pressure (ABP) (Fog 1938, Lassen 1959. The mechanism behind maintaining constant CBF through constriction and dilation of cerebral blood vessels is often referred to as cerebrovascular reactivity (CVR) (Fog 1938, Lassen 1959. Although it should be noted that CVR is the broader term for CA, describing the physiologic process, therefore CVR and CA are not entirely interchangeable since CVR can occur outside the limits of autoregulation. Impairments in CA have been documented in various neuropathological states including stroke (Budohoski et al 2013, Xiong et al 2017, Budohoski and Czosnyka 2018, Czosnyka et al 2020 and traumatic brain injury (TBI) (Czosnyka et al 1997, Sorrentino et al 2012, Donnelly et al 2019, Åkerlund et al 2020, Bennis et al 2020. CA can be visually represented by the Lassen autoregulatory curve, where CBF is plotted against cerebral perfusion pressure (CPP) or mean ABP (MAP), with this curve depicting relatively constant CBF between lower and upper limits of autoregulation (LLA and ULA) (Lassen 1974). When the brain is exposed to pressure-passive flow states, it can lead to hypoperfusion (i.e. ischemia) due to low CPP or MAP below the LLA or hyperperfusion (i.e. hyperemia) due to high CPP or MAP in excess of the ULA, and this can happen in a normal state but in neuropathological states, the region between the LLA and ULA is narrowed (Fog 1938, Lassen 1959, 1974. A significant driver of poor long-term outcomes in various neurological conditions has been attributed to exposure to impaired CA, as suggested in literature (Güiza et al 2015, Donnelly et al 2019, Åkerlund et al 2020, Donnelly et al 2021. Thus, it is becoming more apparent that monitoring CA continuously and accurately at the bedside is ideal for potential early detection, and future intervention, aimed at avoiding states of pressure passive flow.
Since continuous and accurate measures of CBF are not readily available to treating clinicians, the direct measurement of CA is not possible at bedside in humans. Although, we can indirectly measure CA at bedside with the use of surrogate metrics of CA, termed CVR metrics (Czosnyka et al 1997, Zeiler et al 2018b, 2020b. These continuous CVR metrics evaluate the relationship between slow vasogenic fluctuations in CPP/MAP and a surrogate for pulsatile CBF or cerebral blood volume (CBv (Panerai et al 2023)). The most readily studied, with widespread adoption by clinicians at bedside, are the CVR metrics based in the time-domain over the frequencydomain metrics. To obtain the raw physiological signals and derive the surrogate measures for pulsatile CBF/ CBv, invasive, minimally-invasive, and non-invasive modalities can be used such as intracranial pressure (ICP) (Czosnyka et al 1997, Jaeger et al 2006, Zweifel et al 2008, Sorrentino et al 2012, transcranial doppler (TCD) (Czosnyka et al 1997), near-infrared spectroscopy (NIRS) (Zweifel et al 2010a, Zeiler et al 2017c, Chen et al 2020, Sainbhi et al 2022, thermal diffusion flowmetry (TDF) (Rosenthal et al 2011, Dias et al 2015, Highton et al 2015, laser doppler flowmetry (LDF) (Brady et al 2008, Zweifel et al 2010b, Lee et al 2011, Zeiler et al 2018a, 2018b, and brain tissue oxygen (PbtO 2 ) (Jaeger et al 2006, Dengler et al 2013. For more information regarding non-invasive and minimally-invasive modalities, we refer the interested reader to a recently conducted narrative review from our group (Sainbhi et al 2022). CVR metrics are derived as moving Pearson correlation coefficients between slow-wave (i.e. 0.05-0.005 Hz) (Fraser et al 2013, Howells et al 2015 fluctuations in driving pressure for CBF, such as CPP or MAP, and a surrogate for pulsatile CBF/CBv (Lee et al 2009, Brady et al 2010, Zweifel et al 2010a, Zeiler et al 2017a, 2017c, Mathieu et al 2020, Zeiler 2021, Gomez et al 2021a. Currently, the pressure reactivity index (PRx, correlation between ICP and MAP) is the most established method of continuous bedside assessment of CVR, given routine use of invasive ICP monitoring in neurocritical care (Zweifel et al 2008, Czosnyka et al 2009, Copplestone and Welbourne 2018, with pre-clinical literature supporting both ICP and NIRS-based metrics in their ability to measure aspects of the LLA .
Despite increasing literature on the use of these continuous time-domain based CVR measures, there are some critical limitations regarding their use. First, CVR metrics derived from ICP, CPP, TDF, or PbtO 2 currently rely on invasively placed monitors for acquisition of temporally resolved data related to a surrogate of pulsatile CBF (i.e. TDF, PbtO 2 ) or CBv (ICP). This often limits their use to specialized centers, and to a single area of brain in the studies published to date (Czosnyka et al 1997, Steiner et al 2002, Jaeger et al 2006, Zweifel et al 2008, Rosenthal et al 2011, Sorrentino et al 2012, Dengler et al 2013, Dias et al 2015, Howells et al 2015, Lang et al 2015. Second, the non-invasive based indices, derived from TCD or NIRS monitoring, suffer from intra-and interoperator reliability (i.e. TCD), or limited channel capacity (i.e. TCD: bilateral middle cerebral artery insonation Smielewski 2018, Zeiler et al 2018c); NIRS: bifrontal optode placements (Gomez et al 2021a)). Thus, a common narrative between all of these continuous time-domain metrics is the limited spatial resolution in many of the human studies to date. This is despite early work on static and dynamic neuroimaging based techniques of CVR and CBF assessments demonstrating regional heterogeneity in physiologic responses in both states of health and disease (Wintermark et al 2001, Chieregato et al 2003, Aaslid et al 2007, Horsfield et al 2013, Tekes et al 2015, Zeiler et al 2017b, Saito et al 2018, Polinder-Bos et al 2020, Gomez et al 2022, Sainbhi et al 2022.
Thus, a comprehensive understanding of regional disparity in CVR measurements using continuous timedomain indices is required, so that existing knowledge gaps can be identified, and future research directed to address such short comings. The goal of this systematically conducted scoping review is to review human literature that assessed any documented association between continuous CVR between more than one brain region/channel using the same CVR index.

Materials and method
A systematically conducted scoping review of the available literature was conducted using the methodology outlined in the Cochrane Handbook for Systematic Reviews (Cochrane Handbook for Systematic Reviews of Interventions 2021). The data was reported in line with the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) (Page et al 2021), and PRISMA Extension for Scoping Review (Tricco et al 2018). Appendix A of the Supplementary Materials provides the PRISMA checklist. The search strategy and methodology are similar to other scoping reviews published by our group (Hasen et al 2020, Froese et al 2020a, 2020b, Gomez et al 2021b, Batson et al 2022, Gomez et al 2022. The review questions and search strategy were decided upon by the supervisor (FAZ) and the primary author (ASS).

Search questions, population, and inclusion and exclusion criteria
The question posed for this scoping systematic review was: what human literature exists on regional disparities in continuous time-domain CVR between more than one brain region/channel? All studies, either prospective or retrospective, of any size were included.
The primary outcome was the association between measures of continuous time-domain CVR between more than one brain region or channel using the same CVR index. Continuous time-domain CVR measures were defined as those which are moving Pearson (linear) correlation coefficients between slow-wave fluctuations in a driving pressure for CBF (i.e. either MAP or CPP) and a surrogate for pulsatile CBF/CBv, as defined in the existing literature body (Zeiler et al 2018a(Zeiler et al , 2018b. The following cerebral monitoring techniques were considered eligible for the derivation of continuous CVR metrics: ICP, LDF CBF, NIRS, PbtO 2 , TCD or TDF CBF. Appendix B provides a table of the CVR metrics derived from these devices which were of interest in this review.
All studies whether prospective or retrospective, of all sizes, including any human studies that measured time-domain continuous metrics in more than one region or channel using and documented on the association between the regional/hemispheric data were eligible for inclusion in this review. Exclusion criteria were the following: non-English language, animal models, non-continuous CVR assessments, non-pressure based CVR measures (i.e. chemo-reactivity or CO 2 reactivity testing), frequency-domain based continuously updating measures (i.e. transfer function (TF) and related metrics), and non-original works. The frequency-domain measures, such as autoregulatory index (ARI) (Tiecks et al 1995) or TF-ARI (Liu et al 2016), were part of the exclusion criteria because they are not easily computed continuously at the bedside (Zeiler et al 2017b, Gomez et al 2021c. While in TBI patients, ARI has been shown to correlate with time-domain based mean flow index (Mx) (Liu et al 2015), this study has been restricted to those based in the time-domain due to their ease of computation, making them a viable bedside metric. Time-domain based indices covered in this review have received much wider support for clinical adoption in continuous monitoring of patients with a variety of conditions (Czosnyka and Miller 2014, Hawryluk et al 2019, and thus were the main focus for this particular scoping review.
Search strategy MEDLINE, BIOSIS, EMBASE, Global Health, Scopus, and Cochrane Library were searched from inception to the beginning of June 2022 using individualized search strategies for each database. The search strategies for all the databases can be seen in Appendix C of the Supplementary Materials. Finally, the reference lists of reviewed articles on regional disparities in continuous CVR were examined to ensure no references were left out.

Study selection
Using two reviewers (ASS and IM), a 2-step review of all articles returned by our search strategies was performed. First, the reviewers independently screened all titles and abstracts of the returned articles to decide whether they met the inclusion criteria. Second, the full text of the chosen articles was assessed to confirm whether they met the inclusion criteria, and that the primary outcome of regional disparities in continuous CVR was documented. Finally, any discrepancies between the two reviewers were resolved by a third party (FAZ).

Data collection
Data fields included the following: patient characteristics (such as biological sex, age, and relevant disease information), CVR indices measured, regions of brain assessed, primary/secondary outcomes, limitations, CVR values, CVR assessment characteristics, and conclusions regarding continuous indices.

Bias assessment
Given that the goal of this review was to provide a comprehensive scoping overview of the available literature, a formal bias assessment was not conducted.

Statistical analysis
A meta-analysis was not performed in this study because of the heterogeneity of study designs and data, and that the goal of the study was to perform a scoping overview of the available literature.

Results
Search results and study characteristics The results of the search strategy across all databases and other sources are summarized in figure 1. There was a total of 11 478 articles identified from the databases searched. A total of 3031 articles were removed because of duplicated references, leaving 8447 articles to review. By applying the inclusion/exclusion criteria to the title and abstract of these articles, we identified 59 articles that fit these criteria. There were 5 articles added from references sections of articles identified from the first filter, which left 64 articles to review. On applying the inclusion/exclusion criteria to the full-text documents, only 22 were found eligible for inclusion in the systematic review. The articles excluded because they either did not report details around the association between measures of continuous CVR between more than one brain region or channel, were review articles, were non-human literature, or were non-relevant since they reported on non-pressure based CVR measures.
PRISMA, preferred reporting items for systematic reviews and meta-analysis Appendix D gives a general overview of all 22 human studies with primary and secondary outcomes of the study and its limitations (Piechnik et al 1999, Lang et al 2003, Schmidt et al 2003a, 2003b, Soehle et al 2004, Yam et al 2005, Lavinio et al 2007, Joshi et al 2010, Reinhard et al 2003, 2004, 2008, Diedler et al 2011, Haubrich et al 2011, Budohoski et al 2015, Hori et al 2015, Bindra et al 2016, Adatia et al 2020, Meyer et al 2020, Zipfel et al 2020. Table 1 Tables 2, 3 , 4, 5, 6, 7, 8, and 9 describe eight categories of diseases. To measure surrogate for pulsatile CBF/CBv, 5 studies used NIRS (Diedler et al 2011, Hori et al 2015, Bindra et al 2016, Adatia et al 2020, Zipfel et al 2020, and 18 studies used TCD (Piechnik et al 1999, Lang et al 2003, Schmidt et al 2003a, 2003b, Soehle et al 2004, Yam et al 2005, Lavinio et al 2007, Joshi et al 2010, Reinhard et al 2003, 2004, 2008, Haubrich et al 2011, Budohoski et al 2015, Hori et al 2015, Meyer et al 2020 where one study used both (Hori et al 2015). All eligible studies reported on the association between measures of continuous CVR amongst more than one brain region or channel using the same CVR index. CVR indices measured in the studies included: 1 study measured cerebral flow velocity index with ABP (CFVx-a-correlation between ultrasound-tagged NIRS (UT-NIRS) data and MAP) (Hori et al 2015), 3 measured cerebral oximetry index with ABP (COx-a-correlation between rSO 2 and ABP) (Bindra et al 2016, Adatia et al 2020, Zipfel et al 2020, 3 measured diastolic flow index with ABP (Dx-a-correlation between diastolic flow velocity (FV d ) and ABP) (Reinhard et al 2003(Reinhard et al , 2004(Reinhard et al , 2008, 1 measured hemoglobin volume index with ABP (HVx-a-correlation between relative tissue hemoglobin concentration (rTHb) and ABP) (Zipfel et al 2020), 3 measured mean flow index (Mx-correlation between mean flow velocity (FV m ) and CPP) , Schmidt et al 2003a, Lavinio et al 2007, 16 measured Mx with ABP (Mx-a-correlation between FV m and ABP) (Piechnik et al 1999, Lang et al 2003, Schmidt et al 2003b, Soehle et al 2004, Yam et al 2005, Lavinio et al 2007, Joshi et al 2010, Reinhard et al 2003, 2004, 2008, Haubrich et al 2011, Hori et al 2015, Meyer et al 2020, 4 measured systolic flow index with ABP (Sx-a-correlation between FV s and ABP) (Piechnik et al 1999, Reinhard et al 2003, Soehle et al 2004, Budohoski et al 2015, and 1 measured total hemoglobin index with ABP (THx-a-correlation between tissue hemoglobin index (THI) and ABP) (Diedler et al 2011). There were 4 studies with healthy volunteers (Piechnik et al 1999, Schmidt et al 2003b, Yam et al 2005, Reinhard et al 2008, two studies with Cardiac Surgery patients (Joshi et al 2010, Hori et al 2015, three studies with Endarterectomy patients (Reinhard et al 2003, 2004, Zipfel et al 2020, one study with Intracerebral Hemorrhage (ICH) patients (Reinhard et al 2010), two studies with Subarachnoid Hemorrhage (SAH) patients   (Reinhard et al 2008). ABP was monitored non-invasively in all the studies and mean ABP (MAP) was used in the calculation of CVR indices (Piechnik et al 1999, Schmidt et al 2003b, Yam et al 2005, Reinhard et al 2008. All of the studies calculated Mx-a using mean CBFV versus MAP (Piechnik et al 1999, Schmidt et al 2003b, Yam et al 2005, Reinhard et al 2008, while one study also calculated Dx-a using mean CBFV and diastolic ABP (Reinhard et al 2008), and another study calculated Sx-a using mean CBFV and systolic ABP (Piechnik et al 1999). Most of the healthy volunteer studies reported that there were minimal to no difference between Mx-a from left and right sides (Piechnik et al 1999, Schmidt et al 2003b, Yam et al 2005 (p > 0.05 (Piechnik et al 1999), Yam et al 2005) along with one study reporting no difference between Sx from left and right sides (Piechnik et al 1999), and another study reporting no difference between Dx-a and Mx-a calculated via PICA and MCA (Reinhard et al 2008).
The study by Piechnik and colleagues found an increase in Mx-a (>0.4) and Sx-a (>0.2) due to hypercapnia compared to normocapnia in most patients (Piechnik et al 1999). Reinhard and colleagues did not find any influence of age or sex on Dx-a or Mx-a that were calculated by either PICA or MCA. They also stated no major difference was found between cerebellar or cerebral circulation based on the two autoregulatory indices, Dx-a and Mx-a (Reinhard et al 2008). Mx-a threshold between normal and abnormal autoregulation was set as 0.45 by Schmidt and colleagues after inferring from other studies that the threshold assessed with ABP is on average 0.15 greater than the threshold of 0.3 assessed with CPP and they found that 6 healthy volunteers in their study had Mx-a above the threshold either on one side in four volunteers or on both sides in two volunteers (Schmidt et al 2003b).

Cardiac surgery studies
There were only two studies with patients undergoing cardiac surgery that required cardiopulmonary bypass (CPB) using TCD to insonate the MCAs bilaterally as seen in table 3 (Joshi et al 2010, Hori et al 2015, while one study also used UT-NIRS to obtain signals from both sides of the forehead (Hori et al 2015). One of these studies calculated Mx-a and CFVx-a (Hori et al 2015), while the other study only measured Mx-a with a warm control group (arterial inflow temperature 35°C) and two Hypothermia groups (arterial inflow temperature 34°C),   • Mx-a Mx-a: • Insonated MCA bilaterally • Mx-a had high correlation coefficients between sides and low left-right difference • Mx-a L/R avg : 0.37 ± 0.24 (−0.09 to 0.75) ABP, arterial blood pressure; CBFV, cerebral blood flow velocity; CVR, cerebrovascular reactivity; Dx, diastolic flow index; Dx-a, Dx with ABP; FV, flow velocity; FVm, mean FV; FVs, systolic FV; Hz, hertz; MCA, middle cerebral artery; mmHg, millimeter of mercury; Mx, mean flow index; Mx-a, Mx with ABP; NIBP, non-invasive beat-to-beat blood pressure; PICA, posterior inferior cerebellar artery; Sx, systolic flow index; Sx-a, Sx with ABP; TCD, transcranial Doppler. 65 ± 8.8 66 ± 11 64 ± 10 67 ± 13 Prior cerebral vascular event (n, %) 6 (%) Prior carotid endarterectomy (n, %) 2 (3.1%) NR NR NR Prior chronic obstructive pulmonary disease (n, %) 11 (17.2%) 6 (10%) 6 (9%) 2 (17%) Current smoker (n, %) 11 (17.2%) 9 (15%) 7  Joshi et al (2010) Hori et al (2015) Hypothermia group with Mx-a < 0.4 • Insonated MCAs at a depth of between 35 and 52 mm bilaterally • Insonated MCAs at a depth of 35 and 52 mm bilaterally using TCD • Ultrasound-tagged NIRS signals were obtained via adhesive pads attached on the right and left sides of forehead • ABP was monitored invasively from radial artery • ABP was monitored invasively from radial artery • TCD and ABP data were sampled at 60 Hz • Continuous moving Pearson coefficient between MAP and TCD blood flow velocities (Mx-a) and between MAP and UT-NIRS signals (CFVx-a) • Mx-a was calculated between MAP versus TCD blood flow velocities • Mx-a and CFVx-a were calculated using consecutive, paired 10 s averaged values from 300 s duration, incorporating 30 data points for each index Conclusions regarding regional disparity • Statistically significant correlating and agreement between Mx-a and CFVx-a among the subjects (p < 0.001) • During cooling, Mx-a (left, 0.29 ± 0.18; right, 0.28 ± 0.18) was higher than baseline (left, 0.17 ± 0.21; right, 0.17 ± 0.20; p 0.0001) • Mx-a had similar values on both sides • During rewarming phase of CPB, Mx-a (left, 0.40 ± 0.19; right, 0.39 ± 0.19) increased as compared with baseline (p 0.0001) and cooling phase (p 0.0001) indicating impaired CBF autoregulation • CFVx-a had similar values on both sides • After CPB and before wound closure, Mx-a (left, 0.27 ± 0.20; right, 0.28 ± 0.21) was higher than baseline (left, p = 0.0004; right, p = 0.0003), no different than cooling phase of CPB (p = 0.8996) but lower than during rewarming phase (left, p 0.0001; right, p 0.0005) • There was no difference between left (p = 0.2948) and right sided (p = 0.2476) Mx-a between the first and second hour of CPB in the warm control group a Warm control group had arterial inflow maintained at >35°C. b Values reported for warm controls with impaired CBF autoregulation (Mx > 0.4). ABP, arterial blood pressure; CBF, cerebral blood flow; CFVx, cerebral flow velocity index; CFVx-a, CFVx with ABP; CPB, cardiopulmonary bypass; CVR, cerebrovascular reactivity; Hz, hertz; MAP, mean arterial blood pressure; MCA, middle cerebral artery; Mx, mean flow index; Mx-a, Mx with ABP; NIRS, near-infrared spectroscopy; NR, not reported; SD, standard deviation; TCD, transcranial doppler; UT-NIRS, ultrasound-tagged NIRS.  • NIRS signals (rSO 2 and rTHb) were directly measured via bilateral probes attached to the patient's frontal region • Continuous non-invasive ABP was recorded via servocontrolled finger plethysmograph with the patient's right hand positioned at heart level • CBFV measured by insonation of both MCAs through temporal bone window with TCD using 2 MHz transducers attached to a headband • MAP was taken invasively from arterial line • TCD and ABP data were sampled at 100 Hz • Continuous non-invasive ABP was recorded via servocontrolled finger plethysmograph with the patient's right hand positioned at heart level • COx-a and HVx-a were calculated over consecutive paired 10 second averaged values from 300 s epochs samples of MAP versus rSO 2 and rTHb, respectively • Mean Dx-a, Mx-a, and Sx-a were calculated from 20 consecutive 3 s averages of CBFV versus diastolic, mean, and systolic ABP values, respectively • TCD and ABP data were sampled at 100 Hz • Mean Dx-a, Mx-a, and Sx-a were calculated for 1 min periods of 10 min time series from 20 consecutive 3 s averages of CBFV versus diastolic, mean, and systolic ABP values, respectively • Sx-a was not considered for analysis since it came out less reliably • Dx-a (Groups A, B, C: p < 0.001; Group D: p < 0.01), Mx-a (Groups A, B, C: p < 0.001; Group D: p < 0.01), and HVx-a showed significant difference between ipsilateral side of stenosis versus contralateral side • Dx-a and Mx-a clearly showed poor autoregulation values before the procedure compared with contralateral sides • No shunt group showed a significant decrease in HVx-a and COx-a after clamping which indicates intact autoregulation • Correlation coefficient autoregulatory parameters did not significantly differ between symptomatic and asymptomatic stenosis • Patients with stenosis ⩾90% had poorer ipsilateral values for Dx-a and Mx-a than patients with stenosis 75% to 89% (p < 0.05) • Baseline versus Clamping: • Clear side-to-side difference was found in unilateral stenosis ⩾80% for Dx-a and Mx-a but not for Sx-a • Autoregulatory parameters noticeably improved reaching contralateral unaffected sides after recanalization of obstructed ICA • In no shunt group, there was a statistically significant decrease in COx-a ipsilaterally (p = 0.0214) while the contralateral values stayed stable • Dx-a and Mx-a showed the most pronounced difference between 70%-79%, and 80%-89% degree of stenosis groups while the 90%-99% and 100% degree of stenosis groups tended to have poorer values • Ipsilateral degree of autoregulatory improvement was highly significantly related to autoregulatory values before recanalization as shown by correlation coefficient analysis • In shunted group, there was statistically significant increase in ipsilateral COx-a (p = 0.048) prior to shunt insertion but not in the contralateral side.
• Ipsilateral Dx-a and Mx-a proved to be more useful than Sx-a in terms of intergroup differences (groups A-C: p < 0.001; groups A-D: p < 0.01) • Pooled ipsilateral and contralateral data showed statistically significant increase in HVxa in shunt patients (0.05 ± 0.01 versus 0.15 ± 0.02; p < 0.001) compared to no shunt patients (0.073 ± 0.020 versus 0.037 ± 0.014; p = 0.12) • Pooled ipsilateral and contralateral data showed statistically significant increase in COxa in shunt patients (0.18 ± 0.05 versus 0.23 ± 0.01; p = 0.039) compared to no shunt patients (0.17 ± 0.04 versus 0.10 ± 0.01; p = 0.002) Baseline versus post-clamping: • In no shunt group, there was a statistically significant decrease in HVx-a ipsilaterally (p = 0.007) while the values remained stable contralaterally • COx-a showed no statistically significant difference in no shunt group • In shunted group, no significant differences were observed for COx-a or HVx-a ipsilaterally or contralaterally Clamping versus post-clamping: • Ipsilateral and contralateral COx-a and HVx-a remained stable in both no shunt and shunted groups CoW: • When comparing ipsilateral COx-a after clamping, statistically significant decrease of ipsilateral COx-a seen in group with intact CoW (0.18 ± 0.02 versus 0.14 ± 0.02; p = 0.013) as compared to contralateral COx-a or in patients with an impaired CoW • In patients without any anatomical effect, statistically significant decrease in COx-a during ipsilateral clamping (0.19 ± 0.03 versus 0.09 ± 0.02; p < 0.001) compared to contralateral COx-a (0.18 ± 0.01 versus 0.15 ± 0.03; p = 0.22) but no statistical significance difference in other values ABP, arterial blood pressure; CBFV, cerebral blood flow velocity; CEA, carotid endarterectomy; CoW, Circle of Willis; COx, cerebral oxygenation index; COx-a, COx with ABP; CVR, cerebrovascular reactivity; Dx, diastolic flow index; Dxa, Dx with ABP; HVx, hemoglobin volume index; HVx-a, HVx with ABP; ICA, internal carotid artery; MAP, mean arterial pressure; MCA, middle cerebral artery; MHz, megahertz; Mx, mean flow index; Mx-a, Mx with ABP; NIRS, nearinfrared spectroscopy; NR, not reported; rSO 2 , regional oxygen saturation; rTHb, relative tissue hemoglobin concentration; SD, standard deviation; SPAC, stent-protected angioplasty of the carotid artery; Sx, systolic flow index; Sx-a, Sx with ABP. 3.3 ± 1.9 NA Mortality at 90 day (n, %) • CBFV assessed in both middle cerebral arteries with 2 MHz transducers by TCD • ABP continuously and non-invasively recorded via finger plethysmograph • Mx-a was formed from averaged 1 min correlation coefficients calculated from 20 consecutive 3 s averages of CBFV and ABP • Mx-a calculated from MCA sides ipsilateral and contralateral to the ICH • Three measurements of autoregulation were performed Day 1 (12-24 h after ictus), Day 3 (48-72 h) and Day 5 (96-120 h) Conclusions regarding regional disparity • Mean values of Mx-a did not differ significantly across study points • Mean values of Mx-a did not differ when compared with healthy controls of ipsilateral and contralateral sides   • Bitemporal insonation of MCA at depth of 50 mm performed using TCD • ABP monitored either non-invasively or invasively from radial artery • ABP measured invasively from radial artery • Data recorded at 200 Hz frequency • Data sampled at frequency of 50 Hz • Sx-a was calculated using moving, linear correlation coefficient between 10 s averaged values of systolic FV and mean ABP from a 300 s window with a 10 s update frequency • Mx-a and Sx-a were calculated as mean average of Pearson's correlation coefficient calculated among 60 consecutive 5 s averaged values of ABP with mean and systolic FV, respectively Conclusions regarding regional disparity • Worse autoregulation observed ipsilateral to ischemic hemisphere • On vasospasm side, both Mx-a (p = 0.006) and Sx-a (p = 0.044) were higher than on the contralateral side • Correlation between ipsilateral and contralateral Sx-a was lower in patents who developed DCI (p = 0.007) • Mx-a and Sx-a has a correlation (p < 0.001) on the side of vasospasm and on the contralateral side as well • DCI groups had overall worse autoregulation Endarterectomy studies There were three endarterectomy studies found during this review where two studies included patients with severe unilateral stenosis 70% of internal carotid artery (ICA) (Reinhard et al 2003(Reinhard et al , 2004, and included patients with severe carotid artery stenosis 70% (Zipfel et al 2020), as seen in table 4. Two studies insonated MCAs bilaterally using TCD to obtain CBFVs (Reinhard et al 2003(Reinhard et al , 2004, while the other obtained regional oxygen saturation (rSO 2 ) and regional total hemoglobin (rTHb) from NIRS using bilateral probes attached to patient's frontal region (Zipfel et al 2020). ABP was recorded non-invasively in both studies by Reinhard and colleagues (Reinhard et al 2003(Reinhard et al , 2004 while the study by Zipfel and colleagues obtained ABP invasively from an arterial line (Zipfel et al 2020). The CVR indices Dx-a, Mx-a, and Sx-a were calculated in two of the studies by correlating CBFV versus diastolic, mean, and systolic ABP, respectively (Reinhard et al 2003(Reinhard et al , 2004, while the third study calculated COx-a and HVx-a by correlating MAP versus rSO 2 and rTHb (Zipfel et al 2020). These studies showed that Dx-a (Groups A,B,C: p < 0.001 (Reinhard et al 2003); Group D: p < 0.01 (Reinhard et al 2003)), Mx-a (Groups A,B,C: p < 0.001 (Reinhard et al 2003); Group D: p < 0.01 (Reinhard et al 2003)), and HVx-a had significant difference between ipsilateral side of stenosis versus contralateral side (Reinhard et al 2003, 2004, Zipfel et al 2020 while Sx-a (Groups A,B: p < 0.01 (Reinhard et al 2003); Group C: p < 0.05 (Reinhard et al 2003)) either did not show a clear side-to-side difference (Reinhard et al 2003) or was considered less reliable (Reinhard et al 2004), along with COx-a not showing any statistically significant difference between both sides (Zipfel et al 2020).
The study by Reinhard group in 2003 reported that there was a clear side-to-side difference in unilateral stenosis 80% using the Dx-a and Mx-a TCD-based indices, but not for the Sx-a index (Reinhard et al 2003). In  • CBFV was measured in both MCAs via insonation through temporal bone window with 2 MHz TCD • ABP was measured invasively via radial artery • Continuous ABP was measured non-invasively via servo-controlled finger plethysmograph with subject's right hand positioned at heart level • Mx-a was calculated using moving, linear correlation coefficient between 6 s averaged values of mean CBFV and MAP 3 min intervals • Mx-a was calculated as mean average of 1 min Pearson's correlation coefficients calculated among 20 consecutive 3 s averaged values of ABP with mean CBFV Conclusions regarding regional disparity • CA impaired in both groups postinterventional • Not relevant side-to-side differences in Mx-a for both studies • Mx-a was slightly higher on stroke side than contralateral hemisphere in successful recanalization group but ipsilateral side could not be insonated in no recanalization group • Worsening CA later after stroke was indicated by Mx-a being slightly higher in study 2 than in study 1 (p < 0.05) • Contralateral Mx-a showed no significant differences between both groups (p = 0.59) ABP, arterial blood pressure; CA, cerebral autoregulation; CBFV, cerebral blood flow velocity; CVR, cerebrovascular reactivity; ICA, internal carotid artery; MAP, mean arterial pressure; MCA, middle cerebral artery; MHz, megahertz; mmHg, millimeter of mercury; MRI, magnetic resonance imaging; Mx, mean flow index; Mx-a, Mx with ABP; NA, not applicable; NICU, neurological intensive care unit; NIHSS, National Institutes of Health Stroke Scale; NR, not reported; SD, standard deviation; TCD, transcranial doppler.   • Patients with hemispheric asymmetry did not have a worse outcome than patients without hemispheric asymmetry  • Absolute left-right difference in Mx correlated with left-right mean Mx (r = 0.24; p < 0.025), indicating that more autoregulation is impaired, the more asymmetrical it becomes • Significant correlation was seen between the left-right difference in Mx and the midline shift on CT scans (r = −0.42, p = 0.03) which indicates autoregulation is worse on side of brain swelling when there is a midline shift • Mx was higher ipsilateral to contusion than Mx on contralateral side (difference in Mx: 0.16 ± 0.2, p < 0.0035) suggesting autoregulation is worse on side of brain where lesion is present • Absolute left-right difference in Mx was higher in cases of unilateral contusion (0.14 ± 0.18) than in bilateral contusion (0.08 ± 0.1; p < 0.05) which suggests autoregulation is more asymmetrical in cases of unilateral lesion than in bilateral lesion cases • Left-right difference in Mx was significantly higher in patients who died (0.16 ± 0.04) than in those who survived (0.08 ± 0.05; p = 0.04) • Left and right mean Mx were significantly higher in patients who died (0.13 ± 0.05) than who survived (−0.03 ± 0.05; p = 0.002) • Patient outcome was independently correlated with asymmetry of autoregulation (p < 0.0015) • Left-right symmetry accompanies a preserved autoregulation while left-right asymmetry is associated with impaired autoregulation ABP, arterial blood pressure; nABP, non-invasive ABP; CA, cerebral autoregulation; CBFV, cerebral blood flow velocity; CPP, cerebral perfusion pressure; CPP OPT , optimal CPP; CT, computed tomography; CVR, cerebrovascular reactivity; FV, flow velocity; ETCO 2 , end-tidal carbon dioxide; GCS, Glasgow Coma Scale; GOS, Glasgow Outcome Scale; Hz, hertz; ICP, intracranial pressure; ICU, intensive care unit; kPa, kilopascal; mmHg, millimeter of mercury; MCA, middle cerebral artery; Mx, mean flow index; Mx-a, Mx with ABP; NA, not applicable; NR, not reported; PRx, pressure reactivity index; SD, standard deviation; TBI, traumatic brain injury; THI, tissue hemoglobin index; THx, total hemoglobin reactivity index; THx-a, THx with ABP.  • Continuous rSO 2 monitored at a rate of 0.5 Hz using Foresight tissue oximeter using adhesive optodes placed on each side of patient's forehead below the hairline MCA insonation: • ABP was measured invasively from radial or femoral artery catheters at 60 Hz where clinically required • ABP was measured invasively with radial intra-arterial catheter and transducer referenced to the level of the heart • In patients with head injury, MCA insonated daily either on the side of ICP bolt or bilaterally • COx-a was calculated as Pearson correlation coefficient over 10 second intervals between ABP and rSO 2 in a 300 second window • ABP was monitored non-invasively using Finometer photoplethysmograph • In other patients and healthy volunteers, MCA was monitored during clinical test • Hourly COx-a values, taken from mean of all COx-a values obtained through the hour, were obtained separately for each hemisphere • Invasive and non-invasive COx-a were calculated as moving correlation coefficient between 30 consecutive samples of rSO 2 with invasive and non-invasive ABP, respectively, averaged over 10 seconds ICP monitoring: • Global COx-a was calculated as mean of the hourly COx-a values • Hemispheric asymmetry assessed as difference between right and left hemispheres of invasive and noninvasive COx-a values  (2003) • In patients with head injury and SAH, ICP monitored continuously using micro-transducers inserted intraparenchymally into the frontal region • Absolute difference between • In hydrocephalus patients, ICP was monitored using external pressure transducer connected to manometer line ABP monitoring: • In patients with head injury and SAH, ABP was monitored invasively from radial or dorsalis pedis artery • In patients with hydrocephalus carotid artery stenosis and healthy volunteers, ABP was measured non-invasively using finger-cuff CPP was calculated by differencing ICP from ABP Mx/Mx-a was calculated as Pearson's correlation coefficient of 30-60 consecutive samples of mean FV with CPP/ABP Conclusions regarding regional disparity • COx-a asymmetry worsens with each millimeter shift at pineal and septum • Interhemispheric difference of invasive COx-a correlated with interhemispheric difference of non-invasive COx-a (r = 0.81, p < 0.001) • In healthy volunteers, Mx/Mx-a significantly depended on PaCO 2 since at high PaCO 2 , Mx > 0.4 in 86% of volunteers • For each 1 mm shift at pineal and septum, COx-a asymmetry increased by 0.002 ± 0.0005 (p = 0.001) and 0.002 ± 0.0002 (p < 0.001) in univariate analysis, and by 0.009 ± 0.004 (p < 0.001) and 0.005 ± 0.001 (p < 0.001) in multivariate analysis, respectively • In head injury patients with bilateral TCD, Mx/Mx-a was significantly worse on contusion side (p < 0.05) and the side of the brain expansion in patients with midline shift (p < 0.05) • COx-a asymmetry stayed within normal limits of autoregulation in patients without midline shift • Hemispheric differences of Mx/Mx-a were present in most of the patients who died in hospital (p < 0.05) • Beginning of monitoring, COx-a was greater on the injured side but by the end of monitoring, COx-a was greater on the contralateral side • In SAH patients, autoregulation was significantly worse during vasospasm (Mx/Mx-a = 0.46 ± 0.32) than at baseline (Mx/Mx-a = 0.21 ± 0.24; p = 0.021) • COx-a asymmetry did not show any associations with any outcomes (3, 6, and 12 months) • Autoregulation was significantly worse on vasospasm side compared to the contralateral side (p = 0.006) In multivariate analysis:  (2003) • In carotid artery stenosis patients, worse pressure-autoregulation significantly correlated with impaired CO 2 reactivity (p < 0.05) • Regardless of injury location, there was significant relationship between COx asymmetry with midline shift where beta coefficients are 0.013 ± 0.003 (p < 0.001) and 0.005 ± 0.002 (p = 0.004) at pineal, and 0.003 ± 0.001 (p = 0.04) and 0.007 ± 0.003 (p = 0.009) at septum for patients with unilateral injuries and bilateral injuries, respectively • Side-to-side difference in Mx/Mx-a failed to correlate with degree of asymmetry in contrast to data obtained in head injury patients • Significant relationship between COx asymmetry with midline shift in patients with frontal lesions (n = 23) where beta coefficients are 0.005 ± 0.002 (p = 0.002) at pineal and 0.004 ± 0.001 (p < 0.001) at septum • Severe disturbance in autoregulation in presence of bilateral-stenosis compared to unilateral-stenosis as indicated by Mx/Mx-a (p = 0.01) • Significant relationship between COx asymmetry with midline shift in patients without frontal lesions (n = 69) only at septum where beta coefficients are 0.001 ± 0.001 (p = 0.15) at pineal and 0.003 ± 0.0003 (p < 0.001) at septum • In patients with hydrocephalus, Mx/Mx-a significantly correlated with resistance to CSF outflow (r = −0.41; p < 0.03) • This indicates better autoregulation in patients with disturbed CSF circulation ABP, arterial blood pressure; APACHE III, acute physiologic and chronic health evaluation 3 score; CA, cerebral autoregulation; COx, cerebral oximetry index; COx-a, COx with ABP; CPP, cerebral perfusion pressure; CT, computed tomography; CSF, cerebrospinal fluid; CVR, cerebrovascular reactivity; FV, flow velocity; GCS, Glasgow Coma Scale; Hz, hertz; ICH, intracranial hemorrhage; ICP, intracranial pressure; ICU, intensive care unit; IQR, interquartile range; MCA, middle cerebral artery; Mx, mean flow index; NA, not applicable; NIRS, near-infrared spectroscopy; NR, not reported; rSO 2 , regional cerebral oxygen saturation; SAH, subarachnoid hemorrhage; TBI, traumatic brain injury; TCD, transcranial Doppler.
another study by Reinhard and colleagues in 2004, the Dx-a and Mx-a clearly showed poor autoregulation values on ipsilateral side compared to contralateral side (Reinhard et al 2004). Mx-a showed the most pronounced difference between 70%-79% and 80%-89% degree of stenosis groups and Dx-a had a similar trend (Reinhard et al 2003). In one study, patients with 90% degree of stenosis had poorer ipsilateral Dx-a and Mx-a values compared to patients with 75% to 89% degree of stenosis (p < 0.05) (Reinhard et al 2004), and this result was similar to another study by the same first author where the two 90%-99% and 100% groups tended to have poorer values of these two indices (Reinhard et al 2003). There was an autoregulatory improvement on the ipsilateral side after recanalization of obstructed ICA, where the autoregulatory values reached values similar to the contralateral unaffected sides (Reinhard et al 2004).
In the paper that used NIRS-based CVR indices, the no shunt group showed a significant decrease in COx-a and HVx-a after clamping, indicating intact autoregulation, while the shunted group did not, indicating impaired CVR. Compared to baseline, clamping showed significant decrease in ipsilateral COx-a (p = 0.0214) in no shunt group, while ipsilateral COx-a significantly increased prior to shunt insertion in shunted group (p = 0.048), and the pooled ipsilateral and contralateral data showed significant increases in COx-a (p = 0.039) and HVx-a (p < 0.001) in shunted patients (Zipfel et al 2020). Post-clamping compared to baseline showed significant decrease in ipsilateral HVx-a for the no shunt group (p = 0.007) (Zipfel et al 2020). Although postclamping compared to clamping did not result in regional disparity as measured by COx-a and HVx-a, since they remained stable on both ipsilateral and contralateral sides in both the no shunt and shunt groups (Zipfel et al 2020). Also, when comparing ipsilateral COx-a after clamping, a significant decrease was seen in intact Circle of Willis (CoW) group (p = 0.013), and in patients without any anatomical effect, a significant decrease in ipsilateral COx-a was seen (p < 0.001).
Intracerebral hemorrhage (ICH) study As seen in table 5, there was only one study with intracerebral hemorrhage (ICH) patients and healthy controls (Reinhard et al 2010). This study used TCD to acquire CBFV from both MCAs and recorded ABP non-invasively which were used to calculated the Mx-a index by correlating mean ABP with CBFV for both ipsilateral and contralateral sides to the ICH (Reinhard et al 2010). Reinhard and colleagues showed that the mean values of Mx-a did not differ across the study points and also did not differ on ipsilateral and contralateral sides when compared with healthy controls (Reinhard et al 2010). On day 5, higher Mx-a or poor autoregulation was related with lower GCS on both sides (ipsilateral: p < 0.001, contralateral: p = 0.006), presence of ventricular hemorrhage on both sides (ipsilateral: p = 0.011, contralateral: p = 0.018), and lower non-invasive CPP ipsilaterally (p = 0.024) (Reinhard et al 2010). Also, higher ipsilateral Mx-a on day 5 was a significant predictor for poor 90-day outcome (p = 0.013) (Reinhard et al 2010).

Subarachnoid hemorrhage (SAH) studies
For the subarachnoid hemorrhage (SAH) studies, there were only two found during this review that included patients with aneurysmal SAH, as seen in table 6 (Soehle et al 2004, Budohoski et al 2015. Both of these studies insonated MCAs bilaterally to acquire CBFV with TCD (Soehle et al 2004, Budohoski et al 2015, and ABP was acquired purely invasively from radial artery in one study (Soehle et al 2004), while the other study monitored the ABP either non-invasively or invasively from radial artery (Budohoski et al 2015). The CVR indices Mx-a (Soehle et al 2004), and Sx-a (Soehle et al 2004, Budohoski et al 2015 were calculated in these studies by correlating mean ABP with mean and systolic CBFV, respectively. Both of the studies observed worse autoregulation ipsilaterally than on the contralateral side (Soehle et al 2004, Budohoski et al 2015. Soehle and colleagues found that both Mx-a (p = 0.006) and Sx-a (p = 0.044) were higher on vasospasm side than on contralateral side, and both of these indices correlated (p < 0.001) with each other on side of vasospasm, as well as the contralateral side (Soehle et al 2004). Budohoski and colleagues observed worse autoregulation via Sx-a on the ipsilateral side to delayed cerebral ischemia (DCI) and found that there was a lower correlation between ipsilateral and contralateral indices in patients who developed DCI (p = 0.007) (Budohoski et al 2015). Overall, the DCI group had worse autoregulation (p = 0.000 01 for Sx-a DCI versus non-DCI) accompanied with increased interhemispheric difference of autoregulation (p = 0.035), which suggests unilateral autoregulation failure. In the same study, it was found that patients with unfavorable outcome also had worse autoregulation (p = 0.006 for Sx-a favourable versus unfavourable) accompanied with decreased interhemispheric difference of autoregulation (p = 0.027), which suggests bilateral autoregulation failure (Budohoski et al 2015).
Stroke studies There were also only two studies found with ischemic stroke patients where one group had patients with large vessel occlusive stroke (Meyer et al 2020), and the other study looked at patients admitted with acute cerebral ischemia, as seen in table 7 (Reinhard et al 2005). Both studies obtained CBFV bilaterally by insonating MCAs through temporal window with TCD (Reinhard et al 2005, Meyer et al 2020, but one study measured ABP invasively via radial artery (Meyer et al 2020) while the other study measured it non-invasively via finger plethysmograph (Reinhard et al 2005). The Mx-a CVR index was calculated in both studies by correlating CBFV versus MAP (Reinhard et al 2005, Meyer et al 2020. Meyer and colleagues observed that after interventional thrombectomy, CA was impaired in both successful recanalization and no recanalization groups on both ipsilateral and contralateral sides (except ipsilateral side could not be insonated in the no recanalization group) (Meyer et al 2020). While Reinhard and colleagues found no relevant side-to-side differences in Mx-a for their first 48 h study and Day 4-7 after ictus study (Reinhard et al 2005). Mx-a was slightly higher on stroke side than contralateral side, while contralateral Mx-a showed no significant difference between both groups (p = 0.59) in the study by Meyer and colleagues (Meyer et al 2020). In the study by Reinhard and colleagues, CA was observed to be worse later on after stroke, by Mx-a being slightly higher in Day 4-7 study than in the first 48 h study (p < 0.05) (Reinhard et al 2005).
Traumatic brain injury (TBI) studies As given in table 8, five TBI studies were included in this review with varying GCS on admission (Lang et al 2003, Schmidt et al 2003a, Lavinio et al 2007, Diedler et al 2011, Haubrich et al 2011. Only one of these studies measured THI bilaterally over the frontal areas with NIRS (Diedler et al 2011) while the rest of the four studies insonated MCAs bilaterally via TCD to assess CBFV (Lang et al 2003, Schmidt et al 2003a, Lavinio et al 2007, Haubrich et al 2011. In all five TBI studies, ABP was measured invasively from either a radial or femoral artery (Lang et al 2003, Schmidt et al 2003a, Lavinio et al 2007, Diedler et al 2011, Haubrich et al 2011 while three studies measured ICP (Schmidt et al 2003a, Lavinio et al 2007, Haubrich et al 2011, and one study additionally measured ABP non-invasively (Lavinio et al 2007). Mx CVR index was calculated in three studies by correlating CBFV versus CPP (Schmidt et al 2003a, Lavinio et al 2007, Haubrich et al 2011, Mx-a index was calculated in two studies by correlating CBFV versus MAP (Lang et al 2003, Lavinio et al 2007, and only one study calculated THx-a index by correlating THI with MAP (Diedler et al 2011). Two studies reported similar Mx values between both left and right sides (Schmidt et al 2003a, Haubrich et al 2011 while the third study did not report any values for Mx (Lavinio et al 2007). Some patients had Mx-a asymmetry in one study but compared to patients without hemispheric asymmetry, they did not have a worse outcome (Lang et al 2003). With THx-a index, the level of agreement between both sides ranged from high (p < 0.01) to no correlation in patients for the one study calculating the index (Diedler et al 2011).
Although the level of agreement between both sides varied using THx-a index, there was a significant correlation between PRx and THx-a (p < 0.01) when there was good agreement of THx-a on both sides in the study by Diedler and colleagues (Diedler et al 2011). In the study by Lang and colleagues that used the Mx-a index, CA was disturbed in both hemispheres during immediate post-injury phase, but did gradually improve during the ICU course (Lang et al 2003). CA asymmetry between Mx-a and Mx, calculated by subtracting the right side index from the left side index for both indices, had a positive correlation (p < 0.0001) (Lavinio et al 2007). TBI patients in Haubrich and colleagues study that had impaired Mx on both left and right sides showed significant improvement with moderate hypocapnia for both sides (p < 0.001) while patients with normal Mx did not significantly change (p < 0.05) (Haubrich et al 2011). Schmidt and colleagues observed that absolute leftright difference in Mx correlated with left-right mean Mx (p < 0.025) which indicates that the more autoregulation is impaired, the more asymmetrical it becomes (Schmidt et al 2003a). Left-right difference in Mx correlated significantly with midline shift on computed tomography (CT) scans (p = 0.03) which shows autoregulation is worse on the side of brain swelling (Schmidt et al 2003a). They also observed that autoregulation is worse on the lesion side since Mx was higher ipsilaterally than contralaterally (p < 0.0035), and unilateral contusion cases had more asymmetrical autoregulation as compared to bilateral contusion cases (p < 0.05) (Schmidt et al 2003a). Also, it was observed that patients who died had significantly higher left-right difference Mx (p = 0.04), and left and right mean Mx than patients who survived (p = 0.002) (Schmidt et al 2003a). Asymmetry of autoregulation was correlated with patient outcome (p < 0.0015) and hence left-right symmetry seems to go with preserved autoregulation while left-right asymmetry is associated with impaired autoregulation.

Multiple population studies
Multiple population studies was the last group including a total of three studies whose inclusion criteria were acutely comatose patients with GCS 8 (Adatia et al 2020), patients admitted for cardiac arrest, sepsis, stroke, or TBI (Bindra et al 2016), and patients with head injury, SAH, carotid artery stenosis, or hydrocephalus , as described in table 9. Two of these studies monitored rSO 2 bilaterally from the forehead with NIRS (Bindra et al 2016, Adatia et al 2020) while the third study insonated MCA's bilaterally with TCD . Invasive ABP was measured from radial or femoral artery in all three studies , Bindra et al 2016, Adatia et al 2020, but two studies also measured ABP non-invasively , Bindra et al 2016, and one study additionally monitored ICP . Then COx-a index was calculated in two studies by correlating rSO 2 with MAP (Bindra et al 2016, Adatia et al 2020, and the third study calculated Mx/Mx-a by correlating CBFV with CPP and MAP, respectively . Bindra and colleagues observed that the interhemispheric difference of COx-a calculated with invasive MAP correlated with interhemispheric difference of COx-a calculated with non-invasive MAP (p < 0.001) (Bindra et al 2016). While Adatia and colleagues observed that COx-a asymmetry worsened with each millimeter shift at the pineal gland (p = 0.001 and p < 0.001) and septum (p < 0.001 and p < 0.001) in univariate and multivariate analysis, respectively, while COx-a asymmetry stayed within normal limits in patients without midline shift, but COx-a asymmetry did not show any associations with 3, 6, or 12 month outcomes (Adatia et al 2020). Multivariate analysis showed mostly significant relationships between COx-a asymmetry with midline shift at pineal (p < 0.001, p = 0.004, p = 0.002, and p = 0.15), and septum (p = 0.04, p = 0.009, p < 0.001, and p < 0.001) for patients with unilateral injuries, bilateral injuries, frontal lesions, and without frontal lesions, respectively (Adatia et al 2020). Czosnyka and colleagues mentioned that the calculated Mx/Mx-a was significantly worse on contusion side (p < 0.05) or the side with midline shift (p < 0.05) in head injury patients, and most of the patients who died in hospital had hemispheric difference of Mx/Mx-a (p < 0.05). While in SAH patients, autoregulation was significantly worse on vasospasm side compared to contralateral side (p = 0.006) . In contrast to data obtained in head injury patients, the side-to-side difference in Mx/Mxa failed to correlate with degree of asymmetry in carotid artery stenosis patients, while Mx/Mx-a indicated severe disturbance in autoregulation in presence of bilateral-stenosis compared to unilateral-stenosis (p = 0.01) . However, the results for healthy volunteers and hydrocephalus patients did not comment on the regional disparity of Mx/Mx-a .

Discussion
The included human studies reported on the regional disparity using either TCD-based CVR indices (Dx-a (Reinhard et al 2003(Reinhard et al , 2004(Reinhard et al , 2008 al 2011)). The non-invasive nature of TCD and NIRS devices is a big advantage in generating CVR indices with multiple probes as compared to other invasive devices such as ICP, but the temporal resolution of TCD and NIRS needs to be improved to capture the pulse waveform data from the measurements. TCD suffers from intra-and inter-operator reliability, limiting the measurement time, and its spatial resolution is limited due to insonation of bilateral MCA Smielewski 2018, Zeiler et al 2018c). Compared to TCD, NIRS measurement is not as limited, but the commercial NIRS used in clinical settings suffers from limited channel capacity (Gomez et al 2021a), limiting its spatial resolution. Currently there exists research NIRS systems offering increased spatial and temporal resolution by increasing the measurement frequency up to 250 Hz at multiple channels ( , but there is an absence of assessments using NIRS-based CVR metrics, using either commercial of research multichannel NIRS systems, as exemplified by our scoping review. Through the comprehensive evaluation of the human studies on various disease states surrounding regional disparities in continuous CVR between more than one brain region/channel using the same CVR metric, some interesting findings deserve highlighting. First, the available literature only included a handful of CVR metrics, so we can only comment on those select number of metrics that are based on TCD, and NIRS. This may be the case because it is easier to derive multichannel TCD-and NIRS-based indices by adding more channels due to their non-invasive nature compared to ICP-based indices where multiple invasive ICP probes cannot be inserted in different regions of the brain. The handful of studies evaluating TCD-and NIRS-based indices did comment on the regional disparity in various disease states with some studies also adding perturbations to look at the effect it has on the regional disparity. The study by Piechnik and colleagues observed an increase in Mx-a and Sx-a due to hypercapnia compared to normocapnia in most healthy volunteers (Piechnik et al 1999), while the study by Haubrich and colleagues showed that Mx significantly improved on both sides with moderate hypocapnia as compared to normocapnia (Haubrich et al 2011).
Second, there was minimal to no difference found between left and right sides for Dx-a (Reinhard et al 2008), Mx-a (Piechnik et al 1999, Schmidt et al 2003b, Yam et al 2005, Joshi et al 2010, Reinhard et al 2008, Hori et al 2015, Sx-a (Piechnik et al 1999), CFVx-a (Hori et al 2015), in all healthy volunteer, cardiac surgery, and ICH patient studies. This is anticipated for healthy volunteers and cardiac surgery patients, while ICH patients would be expected to have a difference, but it was not the case from the one study found on ICH patients. Most of these studies used TCD-based indices, Dx-a, Mx-a, and Sx-a, while only one of these studies used a NIRS-based CVR index, CFV-x. Although there were only two studies found for cardiac surgery, the study by Hori and colleagues was able to compare the CFVx-a and Mx-a between left and right sides and reported similar values (Hori et al 2015). Interestingly, Joshi and colleagues found that compared to baseline, cooling phase increased Mx-a on both sides, re-warming phase increased Mx-a even more on both sides, and after CPB, Mx-a on both sides came back down to cooling phase levels. While in warm control group, no difference was found in Mx-a between first and second hour of CPB for both sides (Joshi et al 2010). The single ICH study reported that mean values of Mx-a did not differ between the ipsilateral and contralateral sides as compared with healthy controls but mentioned that higher ipsilateral Mx-a on day 5 turned out to be a significant predictor for poor 90-day outcome (Reinhard et al 2010). This highlights the need for future work looking at regional disparity in healthy population and various disease states using NIRS-based indices.
Third, in all endarterectomy, and SAH studies, there was a significant difference reported between ipsilateral side of disease versus contralateral side with the following TCD-based and NIRS-based CVR indices: Dx-a (Reinhard et al 2003(Reinhard et al , 2004, Mx-a (Reinhard et al 2003, 2004, Soehle et al 2004, Budohoski et al 2015, Sx-a (Soehle et al 2004, Budohoski et al 2015, COx (Zipfel et al 2020), and HVx-a (Zipfel et al 2020). This is anticipated since endarterectomy and SAH patients are likely to have asymmetrical conditions. Although a TCD-based index, Sx-a, was reported to either not show a clear side-to-side difference (Reinhard et al 2003), or was considered less reliable (Reinhard et al 2004) in two endarterectomy studies, but this was contrary to two SAH studies that observed worse autoregulation on ipsilateral side to DCI (Budohoski et al 2015) and vasospasm (Soehle et al 2004) as compared to contralateral side with Mx-a and Sx-a. The NIRS-based indices, COx and HVx, showed statistically significant increases in shunted patients from the pooled ipsilateral and contralateral data. Compared to baseline, clamping showed significant decrease in ipsilateral COx-a for no shunt group while it showed an increase in shunted group, but when post-clamping, showed significant decrease in HVx for no shunt group when compared to baseline (Zipfel et al 2020). The Dx-a, Mx-a, and HVx-a indices in endarterectomy studies showed that the ipsilateral side had poor autoregulation values compared to the contralateral side (Reinhard et al 2003, 2004, Zipfel et al 2020, and patients with 90% degree of stenosis had poorer ipsilateral Dx-a and Mx-a values (Reinhard et al 2003(Reinhard et al , 2004. Fourth, there were varying results regarding regional disparity in stroke, TBI, and multiple population studies where some reported no hemispheric asymmetry while others reported worse autoregulation ipsilaterally. One stroke study reported no side-to-side difference with Mx-a in patients admitted with acute cerebral ischemia, although autoregulation was observed to worsen later after stroke (Reinhard et al 2005). While the other stroke study reported that both successful recanalization and no recanalization groups had impaired autoregulation ipsilateral to large vessel occlusive stroke and on the contralateral side, but the ipsilateral side for no recanalization group could not be insonated (Meyer et al 2020). TBI studies reported varying regional disparities where Mx values between both sides were similar according to two studies (Schmidt et al 2003a, Haubrich et al 2011, patients with and without hemispheric asymmetry with Mx-a (Lang et al 2003), level of agreement between both sides ranged from high to no correlation using THx-a (Diedler et al 2011), a NIRS-based index, and one study reported a positive correlation of CA asymmetry between Mx-a and Mx, TCDbased, indices (Lavinio et al 2007). The TBI study that calculated THx-a stated a significant correlation between PRx and THx-a when there was a good agreement of THx-a on both sides (Diedler et al 2011). The TBI study by Schmidt and colleagues made interesting suggestions from their data such as autoregulation becomes more asymmetrical the more autoregulation is impaired, autoregulation is worse on the side of brain swelling and lesion side, and unilateral contusion cases had more asymmetrical autoregulation as compared to bilateral contusion cases (Schmidt et al 2003a). In the multiple population study by Czosnyka and colleagues, they found that Mx/Mx-a was significantly worse on the contusion side (side with midline shift) in head injury patients and on vasospasm side in SAH patients . In the same study the side-to-side difference in Mx/ Mx-a failed to correlate with the degree of asymmetry in carotid artery stenosis patients, which was in contrast to data obtained in head injury patients . Looking at the NIRS-based index of COx-a, Bindra and colleagues stated the interhemispheric difference of COx-a correlated with each other, with the COx-a calculated with invasive and non-invasive MAP (Bindra et al 2016). While Adatia and colleagues observed that COx-a asymmetry worsened with each millimeter shift at pineal and septum, while COx-a asymmetry stayed within normal limits in patients without midline shift, and their multivariate analysis mostly showed significant relationships between COx-a asymmetry with midline shift at pineal and septum for patients with unilateral injuries, bilateral injuries, frontal lesions, and without frontal lesions (Adatia et al 2020).

Limitations
There are some significant limitations that deserve highlighting despite the interesting findings outlined in this scoping review. First, the literature uncovered was very heterogeneous in design which means that there was limited ability to cross-sectionally evaluate the relationship between studies based on various combinations of CVR indices used in each type of disease studies. Second, most studies used a small number of patients in their studies which is likely due to the fact that some patients either do not meet the inclusion criteria or if they do meet the criteria then sometimes the measured signals either do not get properly recorded or end up having too many artifacts. Such small numbers limit the definitiveness of the findings outlined and emphasizes further validation studies. Third, the number of studies per disease is as follows: four healthy volunteer studies, two cardiac surgery studies, three endarterectomy studies, one ICH study, two SAH studies, two stroke studies, five TBI studies, and three multiple population studies. So, comparing between the same disease state was difficult due to lack of studies found commenting on regional disparity of same CVR indices in more than one brain region/channel. Fourth, most of the studies were male dominated and this may not give the full information regarding regional disparity if we assume females hemispheric asymmetry according to disease is different than males. Fifth, all the studies only monitored two brain regions with mostly TCD-based along with some NIRSbased metrics. Most studies only insonated MCAs via transtemporal window with the TCD, while only one study insonated the PICA. No studies were found to insonate arteries such as ophthalmic, basilar, intracranial carotid through other acoustic windows such as transorbital, transforaminal, or submandibular windows (White and Venkatesh 2006, Purkayastha and Sorond 2012, Sainbhi et al 2022. Sixth, the temporal resolution was poor in the studies and most only commented on the mean CVR values for the recording session. TCDs have a practical limitation of less than one hour (Zeiler et al 2017b, Gomez et al 2021c, Sainbhi et al 2022 which explains the poor temporal resolution since most CVR indices measured in the studies were TCD-based. Seventh, the studies reduced the measured indices to grand averages over the short recording period or used limited consecutive points to calculate the CVR indices which restricts the ability to use any high resolution time-series techniques on the indices data. Although this simplifies the analysis, it is at the cost of losing the information encoded in the fluctuations of these indices over the recording period. Lastly, studies that evaluated CA using frequency-domain metrics such as ARI (Tiecks et al 1995), and TF-ARI (Liu et al 2016), were not part of the search strategy since these measures are either non-continuous, computationally complex, or not commonly used, although they may provide value in characterization of CA (Zeiler et al 2017b, Gomez et al 2021c.

Future directions
There are some important areas for future research moving forward. First, validation of above findings is required as some are based on single institution, single study findings, or limited patient numbers. Such work will require multi-institutional collaboration to combine patient data for studies with larger population. Second, as studies uncovered had very heterogeneous patient data, there is a need for more studies to use more homogenous patient datasets based on disease pattern, age, biological sex, and clinical demographics and covers the various disease states. Studying the regional disparity in population of healthy and various disease states is key to understanding how each disease state affects the regional disparity and which indices can better capture this hemispheric asymmetry. Third, spatial resolution needs to increase in order to monitor more brain regions. There is a reason why all the studies found calculated CVR indices based on TCD or NIRS, as these devices can provide a higher spatial resolution of continuous CVR monitoring compared to those derived invasively. The spatial resolution can be increased using multichannel functional NIRS that can measure both oxyhemoglobin and deoxyhemoglobin simultaneously at each channel while being able to remove scalp noise (Chen et al 2020, Gomez et al 2021d, Sainbhi et al 2022. Fourth, the temporal resolution of the data also needs to be increased to obtain pulse waveform data which is possible with NIRS and cumbersome with TCD, due to its practical limit. Obtaining high frequency data is valuable for concurrent derivation of waveform based cerebrovascular parameters that may be of interest in conjunction with CA. These parameters include pressure-flow dynamics, compliance/compensatory reserve, and autonomic function. Fifth, to obtain more comprehensive understanding of physiology, multiple continuous time-domain techniques such as CVR and other raw or derived metrics, should be simultaneously integrated. Sixth, to evaluate temporal changes in regional disparities in disease state as disease progresses, the data needs to be collected during both acute and subacute (long-term) phases and this requires entirely non-invasive, high spatial resolution systems such as NIRS. Seventh, to integrate regional CVR data with patient clinical conditions, patient-reported symptom surveys, and formalized neuropsychological testing toolkits need to be conducted for disease states. Eighth, for healthy control studies, several perturbations such as transient hyperemic response, orthostatic challenge, and neuropsychological tests along with vascular chemo-reactivity are needed, while collecting sufficient numbers of volunteers, to evaluate the impact of age and sex on healthy volunteer reference data. Lastly, evaluating network connectivity between regions is required, once high spatial and high temporal resolution systems are developed.

Conclusion
This literature demonstrates that regional disparity measured by TCD-based and NIRS-based CVR indices may differ according to the disease. In all healthy volunteer, cardiac surgery, and ICH patient studies, Dx-a, Mx-a, Sxa, and CFVx-a indices showed that there was minimal difference between the measured left and right sides. While in all endarterectomy, and SAH studies, there was a significant difference reported between ipsilateral side of disease versus contralateral side using Dx-a, Mx-a, Sx-a, and HVx-a. Also, there were varying results reported regarding regional disparity in stroke, TBI, and multiple population studies. Some studies focused on small number of patients in a specific disease state, so the conclusions drawn from these studies should be taken with caution. There were not many studies evaluating regional disparity using NIRS-based indices, therefore additional research is required to further understand if NIRS-based indices are able to provide better regional disparity information than TCD-based indices.