Crossed Cerebellar Diaschisis Worsens the Clinical Presentation in Acute Large Vessel Occlusion

Introduction: Initial NIHSS in anterior large vessel occlusion (LVO) correlates partially with the hypoperfusion volume. We aimed at assessing the contribution of crossed cerebellar diaschisis (CCD) from the hypoperfused territory on LVO initial clinical deficit. Methods: CCD was retrospectively identified by brain CT perfusion imaging (CTP) in patients with anterior LVO treated by mechanical thrombectomy from January 2017 to July 2021. CCD was defined by CTP parameter alteration in the contralateral cerebellar hemisphere to the LVO. NIHSS, clinical/perfusion variables, and CCD were included in regression models to assess their interrelationships. Results: 206 patients were included. CCD was present in 90 patients (69%). NIHSS scores were higher on admission and at stroke discharge among patients with CCD (17.90 ± 6.1 vs. 11.4 ± 8.4, p < 0.001; 9.6 ± 7.7 vs. 6.6 ± 7.9, p = 0.049; respectively). Patients with a CCD had higher stroke volumes (118.2 ± 60.3 vs. 69.3 ± 59.7, p < 0.001) and lower rate of known atrial fibrillation (22% vs. 41%, p = 0.021). On multivariable logistic regression, CCD independently worsened the initial NIHSS (OR 4.85 [2.37–7.33]; p < 0.001). Conclusion: CCD is found in 69% of LVO on admission CTP, correlates with stroke volumes, and independently worsens initial NIHSS.


Introduction
The National Institutes of Health Stroke Scale (NIHSS) is a graded neurological examination that has become the gold standard for clinical assessments in ischemic stroke trials since the intravenous recombinant tissue plasminogen activator (t-PA) trial for ischemic stroke in 1995 [1,2].The NIHSS is highly reliable across raters from different medical specialties and is now the most widely used deficit rating scale in modern neurology [2].In anterior circulation stroke, admission NIHSS correlates with both stroke volume [3][4][5] and outcome [6] and has become a surrogate to assess initial stroke severity.Change in NIHSS score from baseline to 24 h is part of the evaluation of the efficiency of recanalization therapies [7,8].Cerebrovasc Dis 2023;52:552-559 DOI: 10.1159/000528676However, while there is a correlation between the initial NIHSS and the hypoperfused brain tissue downstream of the occlusion of a cerebral artery, the correlation is only partial, with Spearman's r correlation coefficient ranging from 0.48 to at best 0.68 [3][4][5].Thus, part of the initial NIHSS severity in acute stroke relates to other potential mechanisms than the dysfunction of a brain vascular territory.Diaschisis, a functional impairment of brain areas remote from a cerebral lesion to which it is functionally connected, could contribute to initial NIHSS severity and explain the apparent dichotomy between patients' deficits and the hypoperfused brain area.Diaschisis was first described by Von Monakow in 1914, to explain a functional impairment of structurally normal areas as a result of a remote but connected lesion to the affected area [9].Diaschisis is associated to a matched depression of blood flow and metabolism between the injured brain area and its connected counterparts [10].In stroke, transhemispheric [11,12], thalamic [13], internal capsule [14,15], and cerebellar diaschisis [16][17][18] have been described at both acute and chronic stages.Yet, the relation between diaschisis and initial stroke severity is seldom described.In anterior large vessel occlusion (LVO), the crossed cerebellar diaschisis (CCD) could likely contribute to LVO initial clinical deficit.Indeed, cerebello-cortical connectivity supports motor, cognitive, and perceptual functions [19] thanks to reciprocal connections between the cerebellum, motor, parietal, and prefrontal cortices [20,21].Furthermore, CCD is associated to clinical impairments in both acquired and degenerative neurological diseases [22][23][24].After ischemic stroke, CCD has been reported in several studies using different techniques: single photon emission computerized tomography [22], positron emission tomography (PET) [17,18], MRI perfusion imaging [25], and CT perfusion imaging (CTP) [16,26] with a prevalence ranging from 15% [25] to 58% [17].The bulk of those studies focused on the relation between stroke volumes and CCD [26] or on stroke volumes and long term outcomes [22,25,26].Only one study focused on the role of CCD in acute stroke symptoms and its relation to admission NIHSS using 15 O-water brain PET on 19 patients.This study found a parallel improvement of CCD volume and NIHSS when serially and simultaneously assessed at admission and three and twenty-four hours post-thrombolysis [18].Yet, the results those studies must be balanced by the fact that they included proximal and distal vessel occlusion and were realized before mechanical thrombectomy (MT) was validated for LVO.Here, we aimed at determining, using CTP in a large cohort of anterior circulation LVO who underwent MT, (i) the prevalence of CCD in acute anterior LVO, (ii) the part of the initial clinical deficit explained by CCD, and (iii) the factors associated with the presence of a CCD using logistic regression models.

Study Design and Population
Patients admitted between January 2017 and July 2021 were screened from our stroke registry at Erasmus Hospital in Brussels (Belgium) where all cases of acute ischemic strokes are recorded [27,28].Inclusion criteria were patients (i) aged over eighteen years-old; (ii) presenting with an acute anterior LVO, defined as an occlusion of the ICA (T-type), MCA (M1 or M2 segments), and ACA (A1 or A2 segments); (iii) with brain perfusion CT imaging with reliable analysis by Rapid ® and syngo.via® software, and (iv) who underwent MT.

Cerebral Perfusion Imaging and CCD Identification
Pre-MT imaging included non-contrast brain CT, CT brain/ neck angiography, and CT brain perfusion (CTP).Ischemic core was defined as brain volume with cerebral blood flow (CBF) under 30% of the CBF of the homologous zone in the contralateral hemisphere.Ischemic penumbra was defined as brain volume with time to maximum contrast product arrival exceeded 6 s [28].Those volume were automatically computed with Rapid ® software [29].As Rapid ® perfusion parameter thresholds may not be sensitive enough to identify a CCD, perfusion parameters were also analyzed using a qualitative approach with another software (syngo.via® ) approved in ischemic stroke management.The following perfusion parameters were assessed mean transit time (MTT), time to maximum, time to drain, time to peak, CBF, and cerebral blood volume.CCD was defined as a decrease of CBF and cerebral blood volume and/or an increased MTT, time to peak, time to drain, or T max parameters in the cerebellar hemisphere contralateral to the LVO, compared to the cerebellar hemisphere ipsilateral to the LVO.For subjects to be included, at least three consecutive CTP slices that encompassed the cerebellum were required.Imaging analysis was performed by two independent readers using brain volume CTP interpretation.If a disagreement between readers occurred, results were discussed collegially.

Acute Ischemic Stroke Care and Functional Outcomes
Acute stroke management and care followed ESO guidelines and are detailed in [27,30].MT recanalization was graded using the TICI scale [31].Functional outcomes were evaluated by the NIHSS score on admission and at stroke discharge.
Variable Definition, Group Comparisons, and Regression Models Variable of interest were selected based on previous stroke outcome predictive models: age [32], NIHSS at admission [7,33], ischemic core volume [28], ischemic penumbra volume [34,35],  Color version available online and known atrial fibrillation [36].Those variables, as well as CCD, were chosen as covariates in a linear regression model to predict NIHSS on admission and NIHSS at stroke discharge.The relationship between variables/covariates and outcomes was assessed by the Nagelkerke correlation coefficient.A value of p < 0.05 was considered statistically significant.Then, to assess the relative weight of CCD on outcomes, the other variables were included in the null model as nuisance parameters.A similar approach was used to build a logistic regression model to predict the occurrence of a CCD using the aforementioned variables of interest.Comparisons of selected interest variables between patients with and without CCD were done with a bilateral Student's T test.All statistical analysis was performed using Jasp ® 16.0.

Study Population
296 patients benefited from MT in the period considered.Seventy-five patients lacked CTP on admission; 34 patients had LVO in the posterior circulation, intracranial stenosis, or other abnormalities preventing interpretation; and in 56 cases, CTP was judged either incomplete or uninterpretable, preventing optimal evaluation of CCD.Finally, 131 patients were included (Fig. 1).Main characteristics of the population are detailed in Table 1.
Known atrial fibrillation (KAF) was present in 19 patients out of 87 (22%) in the CCD group while in 17 out of 41 patients (41%) without CCD (p = 0.021).Other demographic and clinical data are illustrated in Table 1. Figure 2a-e and Figure 2a'-d' illustrate a typical CCD.

Logistic Regression Model for the Occurrence of a CCD
The multiple linear regression model including age, NIHSS at admission, ischemic core volume, ischemic penumbra volume, and known atrial fibrillation was statistically significant to predict the occurrence of a CCD treatment (Nagelkerke determination coefficient, R 2 = 0.366, χ 2 (degrees of freedom [dof]:120, 38.4,p < 0.001) and explained 37% of CCD occurrence.In this model, KAF independently accounted for 7% of the outcome variability when age, hypoperfusion volume, core volume, and NIHSS at admission were included in the null model as nuisance parameters (Table 2; Fig. 3).

Linear Regression Model for Admission NIHSS
The multiple linear regression model including age, CCD, ischemic core volume, ischemic penumbra volume, and known atrial fibrillation was statistically significant to predict admission NIHSS (R 2 = 0.39, F [5 dof], 16, p < 0.001) and explained 39% of the variability of the admission NIHSS (R 2 = 0.39).In this model, CCD independently accounted for 7% of the outcome variability when age, hypoperfusion volume, core volume, and KAF were included in the null model as nuisance parameters (Table 3).

Linear Regression Model for NIHSS at Stroke Discharge
The multiple linear regression model including age, NIHSS at admission, ischemic core volume, ischemic penumbra volume, known atrial fibrillation, and CCD was statistically significant to predict NIHSS at stroke discharge (R 2 = 0.23, F [5 dof], 6, p < 0.001) and explained 23% of the variability of the NIHSS 24 h post-treatment (Nagelkerke determination coefficient, R 2 = 0.63).In this model, CCD was not significantly associated to NIHSS at stroke discharge (Table 4).

Discussion
Main findings from this study are that (i) CCD occurs in 69% of patients admitted for acute anterior LVO on CTP, (ii) CCD depends both on admission NIHSS score and hypoperfused volume downstream of the LVO occlusion and shows an inverse relation with known atrial fibrillation, and (iii) CCD independently worsens admission NIHSS.The findings of this study, albeit limited by its monocentric nature, are likely to be generalizable to other populations of LVO treated by thrombectomy.Indeed, our cohort matches closely the characteristics of the large published series on MT in LVO in terms of age, sex, admission NIHSS, and rate of KAF [8,37].
In our cohort, a CCD was observed in 69% of all patients.This proportion parallels the seminal reports on CCD after hemispheric stroke using single photon emission computerized tomography following 133 Xe inhalation and 15 O-water brain PET, where CCD was found in 42% [38] and 58% [13], respectively.However, the CCD rate, we report, is higher than in MRI perfusion weighting imaging (PWI) or MRI blood oxygenation-level dependent cerebrovascular reactivity (BOLD-CVR), where it was described between 15 and 44% [25,39,40], and above the 35% described in the single cohort that used CTP [16,26].The discrepancies between our study and previous works can be reduced to a common denominator.Indeed, CCD is proportional to the total volume of hemispheric hypoperfusion [18,22,26,38,41], a fact corroborated in our study.In the MRI studies, the hypoperfused volume was between 30% [39] and 50% [25] lower than in our cohort, explaining the lower proportion of CCD.Differences in hypoperfusion volume fails, however, to explain why the rate of CCD was lower in the German study that also used CTP to detect CCD: hypoperfusion volume and admission NIHSS were of similar magnitude than in our study, reflecting comparable stroke severity.The lower rate of CCD in that cohort could reflect a selection bias relating to a pre-thrombectomy era, with patients included between 2009 and 2014 [16].Indeed, our population matches the clinical characteristics of the LVO population in the thrombectomy trials and, notably, in terms of the 28% rate of KAF [8], contrasting to the other cohort that used CTP had a population, where patients had a 52% rate of KAF [26].This difference in KAF prevalence may explain the lower rate of CCD they report, as we found in our cohort that KAF is significantly associated to lower likelihood of CCD.The relationship between KAF and CCD may relate to the onset of hypoperfusion: brutal in cardioembolic stroke [42,43] and in more prepared background in LVO due atherosclerotic disease where the LVO often occurs in a context of progressive narrowing of arterial lumen leading to previous relative hypoperfusion.CCD is a process triggered by cerebellar Purkinje cells deactivation from less corticopontine input.The Purkinje cell deactivation leads in turn to lower cerebellar metabolism and blood flow [44], a process that begins at the acute onset of LVO but is maximal after several hours [45].The more progressive onset of hypoperfusion in LVO from non-cardioembolic origin would thus allow a more clear CCD as metabolic and blood flow changes in the cerebellum would have had more time to install.
In our study, CCD accounts for 7% of admission NI-HSS scores, regardless of other factors such as penumbra or core volumes.Therefore, while not associated to a structural lesion, a CCD worsens the initial clinical picture of patients with CCD with by a component potentially fully reversible.The lack of association between CCD and functional outcome is probably explained by two non-exclusive factors: recanalization therapy and lack of sensitivity of the clinical scores.CCD in LVO is associated to brain hemodynamic impairment, vascular steal phenomena due to the recruitment of intracranial arterial collaterals and luxury perfusion [26,40].Recanalization therapies, like in our study, are associated with reversible CCD on follow-up functional imaging [18,46], while CCD may persist up to 5 years in 75% of patients without recanalization therapy [22,47].Thus, recanalization therapy and reversible CCD may explain why CCD clinical effects are obvious only on admission NIHSS.A lack of sensibility of the NIHSS score may also contribute to veil deficits relating to the CCD.Functional outcome scores that are more granular, like the functional ambulation category or the modified Barthel Index, are comparatively more impaired in patients who presented with a CCD, 1 month after the index event [48].The implication of those factors need to be clarified in dedicated longitudinal studies that combine CTP and detailed functional scales.Despite the lack of clear association between CCD and the functional outcome, the presence of CCD is an important parameter to consider in acute stroke workup.First, the analysis of CCD may contribute to more accurate analysis of recanalization therapies efficiency as well as acute functional prognosis.In studies that assessed the benefit of MT+/− intravenous thrombolysis in anterior LVO, the NIHSS improved from 17 to 5 [34,49], in which 10% of the improvement could reflect CCD resolution.This suggests that treatment efficiency studies should control for similar rates of CCD in intervention and non-intervention arms.Second, the presence of CCD is also associated to an increased risk of brain parenchymal hematoma after LVO, and its detection could help orient complication care management [26].

Conclusions
CCD observed with CTP is a frequent phenomenon in anterior LVO strokes treated by MT.CCD is associated with higher stroke volumes and independently worsens admission NIHSS.CCD should be included in the interpretation of admission NIHSS and treatment efficiency, especially when there is a dichotomy between the expected deficit pertaining to the anterior LVO territory and the patient's clinical presentation.Further studies are needed to evaluate the mechanisms and the implications of acute cerebellar dysfunction in the anterior LVO.

Statement of Ethics
The study was reviewed and approved by the Ethics Committee of Erasmus Hospital, Brussels, Belgium (Reference P2018/612).Written informed consent for participation was not required for this study in accordance with the national legislation and the institutional requirement.All methods were performed in accordance with the relevant guidelines and regulations.The research was conducted ethically in accordance with the World Medical Association Declaration of Helsinki.

Fig. 2 .
Fig. 2. Illustration of a typical CCD in left middle cerebral artery (MCA) occlusion.CTP maps show a decrease in CBF (a) and CBV (b) as well an increase in T max (c) in the left hemisphere.There is a CCD characterized by a simultaneous decrease CVF (a') and CBV (b') but also by an increase of Tmax (c') in the contralateral cerebellar hemisphere.MRI-DWI confirms the presence only of a left sylvian stroke (d and d').The left M1 occlusion confirmed by CTA is also illustrated above ([e] with blue arrow).MCA, middle cerebral artery; CTP, brain CT perfusion imaging; CBF, cerebral blood flow; CBV, cerebral blood volume; T max, time to maximum; CCD, crossed cerebellar diaschisis; DWI, diffusion weighted imaging; CTA, computed tomography angiography.

Fig. 3 .
Fig. 3. Illustrates the relationship between known atrial fibrillation (KAF), admission NIHSS, and hypoperfusion volume and the presence of CCD.KAF is related to a lower rate of CCD.CCD is correlated to higher perfusion volume and admission NIHSS.

Table 2 .
Linear regression coefficients for the variables assessed in the prediction model of CCD+ * Significant p value.

Table 3 .
Linear regression coefficients for the variables assessed in the prediction admission NIHSS

Table 4 .
Linear regression coefficients for the variables assessed in the prediction of the NIHSS at stroke discharge Significant p value.