Disability and persistent motor deficits are linked to structural crossed cerebellar diaschisis in chronic stroke

Abstract

Brain imaging has significantly contributed to our understanding of the cerebellum being involved in recovery after non‐cerebellar stroke. Due to its connections with supratentorial brain networks, acute stroke can alter the function and structure of the contralesional cerebellum, known as crossed cerebellar diaschisis (CCD). Data on the spatially precise distribution of structural CCD and their implications for persistent deficits after stroke are notably limited. In this cross‐sectional study, structural MRI and clinical data were analyzed from 32 chronic stroke patients, at least 6 months after the event. We quantified lobule‐specific contralesional atrophy, as a surrogate of structural CCD, in patients and healthy controls. Volumetric data were integrated with clinical scores of disability and motor deficits. Diaschisis‐outcome models were adjusted for the covariables age, lesion volume, and damage to the corticospinal tract. We found that structural CCD was evident for the whole cerebellum, and particularly for lobules V and VI. Lobule VI diaschisis was significantly correlated with clinical scores, that is, volume reductions in contralesional lobule VI were associated with higher levels of disability and motor deficits. Lobule V and the whole cerebellum did not show similar diaschisis‐outcome relationships across the spectrum of the clinical scores. These results provide novel insights into stroke‐related cerebellar plasticity and might thereby promote lobule VI as a key area prone to structural CCD and potentially involved in recovery and residual motor functioning.

Lobule‐specific contralesional atrophy as a surrogate of structural crossed cerebellar diaschisis (CCD) was assessed in a group of 32 chronic stroke patients and healthy controls. Group comparisons revealed structural CCD of the whole cerebellum and particularly of lobules V and VI in stroke patients. The observed volume reduction of lobule VI was associated with higher levels of disability and motor deficits.

1. INTRODUCTION

The cerebellum and its interactions with multiple brain regions of the human motor network have gained an increasing interest in stroke recovery research in recent years. Analyses of focal brain activation (Loubinoux et al., 2007; Rehme et al., 2012; Small et al., 2002), functional (Park et al., 2011; Varkuti et al., 2013; Wang et al., 2010) and structural connectivity (Guder et al., 2020; Schulz et al., 2015) as well as stimulation studies (Chen et al., 2021; Koch et al., 2019; Liao et al., 2021; Rosso et al., 2022) have convergingly indicated that the cerebellum contributes to residual motor functions and recovery after stroke.

Due to its dense connections with multiple cortical and subcortical motor‐ and non‐motor areas of the human brain (Bostan et al., 2013), focal brain lesions, such as caused by non‐cerebellar supratentorial ischemic stroke, can critically impact cerebellar neuronal activity, metabolism, perfusion, and, over time, structural integrity, predominantly affecting the contralesional side of the cerebellum. These alterations have been conceptualized by the term crossed cerebellar diaschisis (CCD) (Baron et al., 1980; Chen et al., 2014; Fan et al., 2013; Gold & Lauritzen, 2002; Pantano et al., 1986). Functional CCD can occur rapidly in the acute stage after stroke. It is presumably caused by the disruption of excitatory corticopontocerebellar fibers (Chan et al., 2017; Pantano et al., 1986), potentially with a frontal motor and premotor origin (Ishihara et al., 1999), which leads to pronounced decreases in Purkinje cell spiking activity, and slight decreases in cerebellar blood flow (Gold & Lauritzen, 2002). One PET study addressed the temporal dynamics in hypometabolism in a rodent stroke model with reperfusion and found that functional, i.e., metabolic CCD occurs from days 7 to 14 after stroke (Joya et al., 2018). More recent studies have highlighted that severe hemodynamic impairment and blood steal phenomena might be involved in early functional CCD as well (von Bieberstein et al., 2021). However, the association of early functional CCD with subsequent outcome remains elusive, given largely negative study results (Abderrakib et al., 2022; Infeld et al., 1995; Komaba et al., 2004; Kunz et al., 2017; Serrati et al., 1994; Wang et al., 2020). For instance, one study found a positive correlation between the presence of early CCD and initial symptom burden. Regarding symptom burden later at hospital discharge, lesion volume (LV) but not CCD was a relevant predictor (Abderrakib et al., 2022). In line, an arterial spin‐labeling MRI study reported a significant association between CCD and initial symptom severity. It also evidenced a significant correlation between CCD and LV (Wang et al., 2020). Another study found that initial functional CCD had a prognostic value for subsequent outcome, but not if the acute‐stage infarct volume was considered in the statistical models (Infeld et al., 1995). A similar influence of LV onto the association between CCD and deficits after stroke in the subacute stage of recovery was noted by other reports (Serrati et al., 1994). Thus, positive correlations between the extent of chronic functional CCD and clinical deficits are likely to be driven by LVs (Szilágyi et al., 2012). These data illustrate that CCD, LVs, initial symptom severity and functional outcome show relevant interdependencies. These call for proper model adjustments in CCD‐outcome modeling, which, if not applied, might also explain negative study results (Kim et al., 1997; Strother et al., 2016).

Functional CCD has been shown to lead to crossed cerebellar atrophy, that is, structural CCD, over time. Herein, cerebellar topography in its connectivity with the brain implies a high spatial specificity not only regarding structural CCD, but also regarding CCD‐outcome relationships. Specifically, the anterior and superior posterior lobe, particularly the lobules IV, V, and VI, are densely connected with the motor cortex. Together with secondary motor representations in the inferior posterior lobule VIII, these lobules contribute to multiple closed‐loop sensorimotor circuits. Posterior lobules VII‐IX as well as Crus I and II of lobule VII are more involved in attentional, cognitive, and executive functions (Guell & Schmahmann, 2020; Hoover & Strick, 1999; Salmi et al., 2010; Stoodley et al., 2021). Well in line with these considerations, stroke‐related structural CCD has been localized in distinct lobules not only of the anterior cerebellar lobe (Abela et al., 2015; Yu et al., 2017) but also for lobules of the superior and inferior posterior cerebellar lobe (Abela et al., 2015; Cai et al., 2016; Wu et al., 2016). However, supportive data for evenly localized CCD‐outcome relationships are strikingly limited (Abela et al., 2015; Cai et al., 2016; Jiang et al., 2017; Sadeghihassanabadi et al., 2022; Wu et al., 2016; Yu et al., 2017). For instance, one study found persistent CCD in lobule VI. However, outcome modeling remained univariate, important influential factors, such as LV or corticospinal tract (CST) damage was not considered, which might explain the missing link with clinical scores (Abela et al., 2015). In acute stroke patients, one study found a significant correlation between cerebellar anterior lobe volume loss and impaired clinical improvement over 6 months, which remained stable after correction for cofactors such as baseline LV (Yu et al., 2017). Another study found a progressive atrophy in lobule VII over 1 year after stroke. Despite the adjustment for the infarct volume, a correlation with clinical scores was not detected (Cai et al., 2016).

Collectively, to what extent specific regions of the cerebellum might be subject to structural atrophy after stroke, i.e., might undergo structural CCD, is not fully understood. Even more importantly, the link between lobule‐specific CCD and measures of disability or persistent motor deficits remains vague, available data are inconsistent. Many previous studies have not incorporated sufficient modeling adjustment for important influential factors, such as age, LV, or the structural state of important corticofugal motor pathways like the CST. Also, outcome scores were limited to measures of global disability and were often not specifically addressing the motor domain (Abderrakib et al., 2022; Infeld et al., 1995; Kunz et al., 2017; Serrati et al., 1994; Szilágyi et al., 2012; Wang et al., 2020; Yu et al., 2017). Few studies used scores to assess motor impairment more directly (Abela et al., 2015; Cai et al., 2016; Kim et al., 1997; Strother et al., 2016; Wu et al., 2016). Hence, the evaluation of CCD‐outcome relationships for a broad spectrum of clinical aspects from global disability to motor deficits seems to be warranted. The aim of the present study was to further address these important questions by (1) quantifying the amount of structural lobule‐specific CCD in a well‐characterized cohort of chronic stroke patients in comparison with healthy controls, and (2) relating the extent of lobule‐specific CCD to clinical scores covering global disability, functional upper limb abilities, and gross and fine motor skills of the hand. CCD‐outcome inferences should be corrected for age, LV, and CST damage. We hypothesized that particularly volumes of lobules related to motor and cognitive functioning, such as lobules I‐VI or VIII, would show stroke‐related structural CCD and significant structure‐outcome relationships.

2. MATERIALS AND METHODS

2.1. Cohort and clinical data

Thirty‐two first‐ever stroke patients with supratentorial strokes in the chronic stage of recovery (at least 6 months post‐stroke) were included in this retrospective analysis. Inclusion criteria were: first‐ever unilateral ischemic stroke, residual upper extremity motor deficit as operationalized by an upper extremity Fugl‐Meyer (UEFM) score ≤65, no history of previous neurological or psychiatric illness, age ≥18 years. Thirty‐two healthy controls of similar age and sex served as a control group. Clinical tests included the UEFM test, the modified Rankin Scale (MRS) as a measure of global disability, the nine‐hole‐peg‐test (NHP) and the maximum grip strength (GS) for fine and gross hand motor output, respectively. NHP and GS values are given as ratios (affected/unaffected hand) with higher values indicating better motor functions. Data of 11 patients were part of a previous analysis of prefrontal motor connections after stroke (Schulz et al., 2019). All participants provided informed consent according to the Declaration of Helsinki. The study was approved by the ethics committee of the Chamber of Physician Hamburg.

2.2. Brain imaging

A 3 T Skyra MRI scanner (Siemens, Erlangen, Germany) equipped with a 32‐channel head coil was used to obtain structural T1‐weighted images applying a three‐dimensional magnetization‐prepared rapid gradient echo sequence (repetition time [TR] = 2500 ms, echo time [TE] = 2.12 ms, flip angle 9°, 256 coronal slices with a voxel size of 0.8 × 0.8 × 0.9 mm³, field of view [FOV] = 240 mm). For cerebellum lobule segmentation and volumetric analysis, the software volBrain (Manjón & Coupé, 2016) and the CERES pipeline (Romero et al., 2017) were used. The following processing steps were conducted: de‐noising, inhomogeneity correction, linear registration to Montreal Institute of Neurology (MNI) space, cropping of the cerebellum, nonlinear registration to an MNI cerebellum template, intensity normalization, and a nonlinear registration to a subject‐specific library. Cerebellar structures were assigned according to published definitions (Park et al., 2014). Volumes were estimated for the cerebellum and 12 cerebellar regions, including lobules I‐II, III, IV, V, VI, crus I, and crus II (VII), VIIB, VIIIA, VIIIB, IX, and X. This workflow has been previously used in a study on brain reserve capacity after stroke (Sadeghihassanabadi et al., 2022). In addition to these volumetric analyses, the imaging data also included diffusion‐weighted images which were used to integrate information on the microstructural damage of the CST, in line with a previous study (Schulz et al., 2019). The diffusion data consisted of 75 axial slices with gradients (b = 1500 s/mm²) applied along 64 noncollinear directions (TR = 10,000 ms; TE = 82 ms; FOV = 256 × 204; voxel size 2.0 × 2.0 × 2.0 mm³). CST damage was quantified by proportional fractional anisotropy (FA, expressed as the ratio ipsilesional [IL]/contralesional [CL]) at the level from the mesencephalon to the cerebral peduncle (MNI: z = −25 to z = −20). CST FA estimation was based on CST templates originating from the primary motor hand area available from a previous report which also includes details on this analysis (Schulz et al., 2015). LVs were calculated based on manually delineated lesion masks on the individual T1‐weighted images. B₀‐weighted diffusion images were considered to guide lesion masking.

2.3. Statistics

R version 4.2.1 was used for statistical analyses. Volumes of the cerebellum and 12 specific lobules were estimated for the IL and CL side. For group comparison, the distribution of lesion locations (right or left hemispheres) in the patients was adopted for the control group leading to equal number of controls with left and right hemispheric pseudo‐lesions, in line with several previous reports (Graterol Pérez et al., 2022; Schulz et al., 2015). Structural CCD was operationalized and computed as the ratio between CL and IL sides (abbreviated as CCD). Only this measure was further analyzed to correct volumetric findings for differences in individual total cerebellar volume, respectively. Group comparison was carried out by linear regression modeling with the CCD as the dependent variable and the group as the independent variable of interest. For CCD‐outcome inference, separate ordinal (MRS) and multiple linear regression models (UEFM, NHP, GS) were computed. The clinical score was treated as the dependent variable, age, LV (log‐transformed, residualized against CCD) and proportional CST FA (CST FA, residualized against CCD) were included as additional independent variables to adjust the target effects. For MRS and ordinal logistic regression, CCD was considered as a binarized variable (more vs. less CCD) based on a median split of the patients. Resulting odds ratios (OR) below 1 would indicate a lower risk of rising one level in MRS for patients with less CCD. Linear residualization was applied to allow for collinearity between structural CCD, LV, and CST damage, similarly to previous studies (Rojas Albert et al., 2022; Sadeghihassanabadi et al., 2022). Statistical significance was adjusted for multiple testing by applying a false discovery rate (FDR) correction, stated as p _FDR. The robustness of the modeling results was further tested by means of a leave‐one‐out model analysis. Statistical significance was assumed at a p‐value <.05.

3. RESULTS

3.1. Clinical and demographic data

Clinical and demographic data are summarized in Table 1. The patient cohort consisted of 32 stroke patients (20 males, 23 right‐sided strokes, 3 left‐handed, mean age 63 years). Then, 32 healthy controls of similar age and sex served as the control group (20 males, 2 both‐handed, mean age 64 years). On average, the patients exhibited moderate persistent clinical deficits with large intersubject variability. Median MRS was 1 (range 0–4, interquartile range 1), mean UEFM score was 48 (range 5–65) and mean proportional NHP and GS performance was 0.5 (0.0–1.1) and 0.6 (0.0–1.1), respectively. Patients were included on average 38 months after stroke (6–215); lesions were variable in location and size (Table 1).

TABLE 1.

Demographic and clinical data.

ID	Sex	Age	DoHe	AfHe	TAS	Stroke location	LV	CST FA	MRS	UEFM	NHP	GS
1	m	69	L	R	97	BG, PCA	52.0	0.9	2	47	0.0	0.5
2	m	74	L	L	33	BG	1.0	1.0	2	46	0.0	0.5
3	f	48	L	R	71	MCA	63.0	0.9	1	54	0.3	0.7
4	m	51	L	R	80	BG	3.9	0.7	3	37	0.0	0.3
5	m	58	L	R	69	MCA	211.2	0.8	2	31	0.0	0.3
6	m	57	L	L	20	BG	1.3	1.0	0	60	0.9	0.8
7	m	71	L	R	27	BG	1.3	0.8	2	49	0.4	0.7
8	f	72	L	R	36	BG, MCA	23.4	0.8	2	39	0.0	0.5
9	m	62	L	L	28	PCG	0.3	1.0	0	63	1.0	1.0
10	m	75	L	R	15	MCA	59.9	0.9	2	46	0.0	0.4
11	m	58	L	L	26	PCG	1.1	1.0	1	59	0.7	1.0
12	f	72	L	L	169	BG, MCA	11.3	1.0	1	61	0.9	1.0
13	f	46	L	L	16	MCA	56.5	1.1	1	61	0.8	1.1
14	f	50	L	L	12	BG	6.8	0.6	3	16	‐	0.4
15	m	56	L	R	13	BG	2.1	0.8	3	6	0.0	‐
16	f	70	L	R	13	BG, MES	1.2	0.6	4	5	‐	‐
17	m	66	L	R	14	MCA	8.2	0.9	1	64	0.7	0.8
18	f	64	L	R	71	MCA	35.9	0.9	1	60	1.1	0.7
19	f	66	R	R	23	BG	0.1	0.9	1	55	1.0	0.6
20	m	65	R	L	12	BG	1.7	0.7	3	13	0.0	0.0
21	m	64	L	L	12	BG	0.3	1.0	1	60	0.9	0.9
22	f	71	L	R	12	MCA	21.3	0.9	1	65	0.7	0.7
23	m	64	L	R	13	BG	0.4	0.9	1	60	0.7	0.4
24	m	64	R	R	215	MCA, ACA	74.3	0.9	1	38	0.0	0.3
25	m	59	L	L	44	BG	3.3	0.9	1	59	0.6	0.8
26	m	51	L	R	12	MCA	19.7	0.9	1	64	0.9	0.9
27	f	86	L	R	13	MCA	26.5	0.8	3	50	0.0	0.4
28	m	62	L	R	6	BG	1.1	0.7	2	59	0.7	0.6
29	f	57	L	R	22	MCA, PCA	70.5	0.8	2	30	0.0	0.3
30	m	49	L	R	8	BG	10.1	0.8	1	60	0.6	0.6
31	f	48	L	R	12	BG, PCG	1.1	1.0	1	64	0.6	0.8
32	m	81	L	L	12	BG	0.3	0.9	1	63	0.8	0.9
Stroke	m: 20	63 ± 9.9	L:29	L:11	38 ± 47.1	‐	24.1 ± 41.6	0.9 ± 0.1	1 a IQR 1	48.3 ± 17.8	0.5 ± 0.4	0.6 ± 0.3
HC	m: 20	64 ± 7.3	L:30	‐	‐	‐

Note: Description of the study cohort including 32 supratentorial stroke patients as well as 32 age‐matched HCs is provided along with clinical scores indicating motor performance in the stroke patients. Clinical scores included mRS, UEFM, NHP (in pegs per second), and GS. GS and NHP values are given as ratio (affected/unaffected hand) with higher values indicating better motor functions. CST FA values are measured at the level of the cerebral peduncles and are given as ratio values (affected/unaffected side) with lower values indication more residual damage to the CST. Sex (m = male, f = female) and age (in years), dominant (DoHe) and affected (AfHe) hemisphere (L = left, R = right), TAS (in months), stroke location (MCA, PCA, ACA, PCG, BG, upper MES), LV (in cm³). Averaged group values are given as mean ± standard deviation (SD).

Abbreviations: ACA, anterior infarct; BG, basal ganglia; CST, corticospinal tract; FA, fractional anisotropy; GS, grip strength; HC, healthy control; LV, lesion volume; MCA, media infarct; MES, mesencephalon; mRS, modified Rankin Scale; NHP, nine‐hole‐peg performance; PCA, posterior infarct; PCG, precentral gyrus; TAS, time after stroke; UEFM, Fugl‐Meyer assessment of the upper extremity.

^a Median MRS with interquartile range (IQR).

3.2. Structural CCD after supratentorial, non‐cerebellar stroke

Group comparisons revealed a significant structural CCD for the whole cerebellum with lower CCD in strokes compared to healthy controls (p _FDR < .001). As hypothesized, structural CCD of the whole cerebellum was significantly correlated with higher LV (β −.45, p < .001) and CST damage (β .44, p < .01). Regarding specific lobules, localized structural CCD was evident for lobules V and VI (p _FDR = .04). Table 2 exhibits all results for the group comparisons between patients and healthy controls.

TABLE 2.

Absolut and proportional cerebellar and lobule volumes.

Region	IL		CL		CCD (CL/IL)		p FDR
	Stroke	HC	Stroke	HC	Stroke	HC
Cerebellum	62.32 ± 7.15	61.8 ± 6.94	59.91 ± 7.17	61.61 ± 7.23	0.96 ± 0.04	1.00 ± 0.02	<.001*
Lobule I_II	0.07 ± 0.02	0.07 ± 0.03	0.06 ± 0.02	0.06 ± 0.03	0.93 ± 0.17	0.92 ± 0.21	1.00
Lobule III	0.74 ± 0.19	0.73 ± 0.17	0.67 ± 0.17	0.71 ± 0.18	0.91 ± 0.13	0.97 ± 0.15	.68
Lobule IV	2.25 ± 0.29	2.17 ± 0.36	2.22 ± 0.35	2.18 ± 0.51	0.99 ± 0.13	1.01 ± 0.17	1.00
Lobule V	3.99 ± 0.64	3.83 ± 0.63	3.76 ± 0.67	3.97 ± 0.72	0.95 ± 0.12	1.04 ± 0.14	.04*
Lobule VI	8.62 ± 1.34	8.71 ± 1.11	8.21 ± 1.32	8.81 ± 1,11	0.96 ± 0.08	1.02 ± 0.08	.04*
Crus I	12.46 ± 1.65	12.59 ± 2.12	11.96 ± 1.88	12.23 ± 1.89	0.96 ± 0.07	0.98 ± 0.09	1.00
Crus II	8.05 ± 1.38	8.34 ± 1.61	7.78 ± 1.21	8.01 ± 1.62	0.98 ± 0.12	0.97 ± 0.12	1.00
Lobule VIIB	4.55 ± 0.77	4.49 ± 0.86	4.15 ± 0.65	4.49 ± 0.84	0.92 ± 0.12	1.01 ± 0.14	.08
Lobule VIIIA	5.84 ± 1.13	5.53 ± 0.89	5.54 ± 1.26	5.75 ± 1.29	0.95 ± 0.12	1.04 ± 0.16	.11
Lobule VIIIB	4.09 ± 0.70	3.9 ± 0.74	4.26 ± 1.03	3.81 ± 0.55	1.04 ± 0.16	0.99 ± 0.10	1.00
Lobule IX	3.5 ± 0.83	3.26 ± 0.70	3.43 ± 0.77	3.29 ± 0.73	0.99 ± 0.11	1.01 ± 0.07	1.00
Lobule X	0.61 ± 0.07	0.6 ± 0.09	0.62 ± 0.08	0.59 ± 0.08	1.01 ± 0.12	0.99 ± 0.07	1.00

Note: Group comparison between chronic stroke patients and HCs of structural crossed CCD of the whole cerebellum and the cerebellar lobules. Cerebellar and lobule volumes are given for both IL and CL sides (in cm³). CCD is assessed via the proportional value between CL and IL volumes. Results are displayed as mean ± standard deviation (SD). p‐Values are given (FDR‐corrected for 13 comparisons).

Abbreviations: CCD, crossed cerebellar diaschisis; CL, contralesional; FDR, false discovery rate; HCs, healthy controls; IL, ipsilesional.

3.3. Relationship between structural CCD and clinical scores

Regression modeling revealed a significant association between localized structural CCD in lobule VI for all four clinical scores tested (all p _FDR < .05, Table 3, Figure 1). Specifically, more pronounced volume reduction in lobule VI, that is, lower CCD values, correlated with higher MRS scores and lower UEFM, NHP, and GS scores. Compared to simplified ordinal/linear regression models in which lobule VI CCD was omitted, its inclusion explained an additional 6% of variance in variability in MRS, 11% in variability UEFM and GS, and 8% in variability in NHP, respectively. Increasing CST damage as operationalized by reduced proportional CST FA values was positively correlated with more severe deficits. Details of the final models are summarized in Supplementary Table 1. Structural CCD of the whole cerebellum showed a significant association with MRS only, otherwise, both cerebellar and lobule V CCD did not show consistent correlations across the tested clinical scores.

TABLE 3.

Relationship between structural CCD and clinical scores.

Region	MRS		UEFM		NHP		GS
	OR	p FDR	β	p FDR	β	p FDR	β	p FDR
Cerebellum	0.06	.014*	.32	.192	.39	.139	.31	.192
Lobule V	0.65	1.000	.00	1.000	.15	1.000	.07	1.000
Lobule VI	0.07	.015*	.42	.019*	.45	.023*	.44	.010*

Note: Results of ordinal (MRS) and multiple linear (UEFM, NHP, GS) regression models relating proportional cerebellar and lobule volumes (ratio contralesional/ipsilesional) to clinical scores in chronic stroke patients (corrected for nuisance variables age, lesion volume, and CST damage). Estimated OR or standardized coefficients (β) are given for the three regions exhibiting significant structural CCD. Results are stable after leave‐one‐out model analyses. p‐Values are FDR‐corrected for 12 tests across all scores.

Abbreviations: CCD, crossed cerebellar diaschisis; CST, corticospinal tract; GS, grip strength; MRS, modified Rankin Scale; NHP, nine‐hole‐peg test performance; OR, odds ratios; UEFM, upper extremity Fugl‐Meyer score.

FIGURE 1.

Outcome correlation with lobule VI structural crossed cerebellar diaschisis (CCD). Modified Rankin Scale (MRS) distribution is illustrated by stacked histograms for more and less lobule VI structural CCD after median split dichotomization (see Section 2). For upper extremity Fugl‐Meyer score (UEFM), proportional nine‐hole‐peg test performance (NHP), and proportional grip strength (GS) adjusted effect plots are given for linear regression analyses with lobule VI structural CCD with linear fit (gray line), 95% confidence intervals (shaded) and individual point estimates. Asterisks indicate significant CCD‐outcome associations.

4. DISCUSSION

There were two main findings in the present study. First, CCD was evident for the whole cerebellum and particularly for lobules V and VI in chronic stroke patients. Second, lobule VI CCD was significantly correlated with clinical scores. Specifically, more pronounced volume reduction in contralesional lobule VI was associated with a higher level of global disability, functional upper limb abilities, and gross and fine motor skills of the hand. This association was independent of age, LV, and CST damage. Lobule V and the whole cerebellum did not show a similar relationship across the spectrum of clinical scores.

The analyses revealed a significant structural CCD not only regarding the whole cerebellum, but more importantly, for distinct cerebellar lobules V and VI of the posterior part of the anterior and anterior part of the superior posterior cerebellar lobes. Regional CCD in lobule VI has been also reported by one observational longitudinal study over 9 months after stroke which supports the present cross‐sectional results and indicates that lobule VI CCD progresses after stroke and persists over time. Though showing lobule VI atrophy, that study has not detected any significant association with motor recovery (Abela et al., 2015). Our results expand these findings and demonstrate that in the chronic stage, the extent of lobule VI CCD is significantly linked to global disability and persistent motor deficits. Interestingly, the patient cohorts were very similar regarding supratentorial lesion locations, LV and MRS at follow‐up. However, the statistical approaches, particularly the adjustment for important cofactors, such as LV and CST damage, were different between the analyses. Together with the study designs, a longitudinal assessment within the first year versus a cross‐sectional analysis after 38 months after stroke in the present study, this might explain the discrepancy in CCD‐outcome results.

In sum, our data further underpin that lobule VI might be a cerebellar key area both prone to stroke‐related structural CCD and potentially involved in recovery and residual motor functioning. Tracing studies have shown that lobules IV, V, and VI are linked with the motor cortex (Hoover & Strick, 1999). Tractography in humans have revealed reciprocal cortico‐cerebellar tracts between lobules V and VI and the primary motor cortex and the dorsal premotor cortex (Salmi et al., 2010). In healthy aging, regional white matter volumes in lobules IV, V, and VI have been associated with motor functions (Koppelmans et al., 2017). Following the emerging concepts of brain reserve capacity for stroke recovery, one recent study has investigated the structural state of the cerebellum directly after the event in a cohort of severely impaired acute stroke patients. Importantly, this study did not assess structural CCD early after stroke, but it aimed to investigate to what extent intersubject variability in cerebellar anatomy might explain outcome variability. It found that larger volumes in lobule VI, IV, and VIIIB were positively correlated with better outcome after 3–6 months. This might suggest that better structural reserve in lobule VI at baseline might attenuate the negative effects of subsequently occurring structural CCD (Sadeghihassanabadi et al., 2022).

It was an interesting finding that lobule VI, but not lobule V showed a significant association with behavior in our patients, albeit both lobules exhibited very similar magnitudes of CCD. The question arises why the relationships with outcome were so different? In fact, a strikingly similar finding has been also observed in a study in severely impaired stroke patients (Sadeghihassanabadi et al., 2022). In that study, baseline lobule V lost statistically significance in outcome models after adjustment for the initial deficit. On the one hand, it has been discussed that lobule VI was also involved in cognitive processes such as attentional and executive processing and working memory tasks with functional connectivity with parieto‐frontal cortical brain networks (Stoodley et al., 2021; Stoodley & Schmahmann, 2010). The upregulation in cognition‐related cortical networks, for example, explained by attentional processes to motor performance or by motor learning strategies, has been already reported after stroke (Li et al., 2016; Marshall et al., 2009; Ward et al., 2003). Interestingly, meta‐analyses in motor learning have revealed that lobules V and VI were involved, but only lobule VI activation remained stable when rather simple motor execution demands were regressed out (Hardwick et al., 2013). Hence, and on a speculative note, lobule VI, spared from structural CCD, might therefore provide larger residual structural and functional reserve for motor learning strategies, contributing to relearning of lost motor functions after stroke. Another reason for the discrepancy between lobules VI and V might lie in the selection of motor tasks and clinical scores. For instance, tactile stimulation of the hand leads to activation in lobule V (Stoodley et al., 2021); hence, sensorimotor integration tasks could prove a functional importance of CCD in lobule V after stroke. In turn, this also indicates that the present results should not be interpreted in the way that other lobules do not contribute to recovered motor functions after stroke.

Collectively, the involvement of lobule VI in the motor and cognitive domain might explain why the extent of CCD in lobule VI, but not in lobule V, exhibited significant correlations with multiple clinical scores in the present cohort. Likewise, this multidomain involvement of lobule VI in first motor and first non‐motor representations (Guell & Schmahmann, 2020) might also explain why previous longitudinal studies showing atrophy, for example, in lobule VII, have not found significant CCD‐outcome associations for this lobule (Abela et al., 2015; Cai et al., 2016; Wu et al., 2016; Yu et al., 2017). Finally, the present largely lobule‐specific outcome findings for lobule VI also indicate that, in general, the whole cerebellum undergoes structural CCD after stroke, but only diaschisis in distinct lobules associates with clinical scores which relate not only to global disability, operationalized by the MRS score, but also to upper limb, gross or fine hand motor deficits. In our cohort, CCD of the whole cerebellum was only linked with global disability. In agreement, our previous report showed that baseline volume of lobule VI, assessed early after severe stroke, but not of the whole cerebellum, was able to inform about subsequent recovery (Sadeghihassanabadi et al., 2022). Ultimately and again theoretically, these data might suggest that innovative cerebellar brain stimulation paradigms (Wessel & Hummel, 2018) might be beneficial to promote recovery after supratentorial, non‐cerebellar stroke when targeting lobule VI as a cerebellar key area involved in important cognitive and motor function in health and after stroke. However, as only few broad motor functions were assessed in the current study, and the amount of explained variance was rather small, treatment effects should not be overestimated. Nevertheless, when applied early after stroke, and probably in combination with cortical stimulations and structured multi‐session training paradigms, it seems warranted to evaluate such novel protocols prospectively in well‐powered clinical trials.

There are several important limitations to note. First, the sample size was relatively small. However, the analyses were biased towards robust findings by correction for multiple testing and a leave‐one‐out model analysis to exclude influential points. This approach leads to high sensitivity at the cost of reduced specificity. Hence, we cannot exclude that also other cerebellar regions might carry important information to explain outcome variability in chronic stroke patients. Second, this was a cross‐sectional study in chronic stroke patients. Albeit our results are supported by previous longitudinal data from the acute to the chronic stage (Abela et al., 2015), we cannot fully rule out that pre‐stroke group differences might have biased the volumetric findings. Third, time after stroke was variable with few patients included after more than 2 years. As a post hoc analysis, we recomputed the winning correlational models for lobule VI while excluding patients with more than 24 months after stroke. CCD‐outcome associations remained stable (all p < .05). Fourth, lesion masking was based on T1‐weighted images. As high‐resolution 3D T2/FLAIR‐weighted images were not available and, additionally, chronic shrinkage and atrophy processes are likely to influence lesion morphology, LVs might be underestimated. This bias, particularly in small lesions, might influence the present structure‐outcome relationships. Fifth, as behavioral data were not available for the healthy participants, it remains a question to whether the present CCD‐outcome association is specific for stroke or whether normal aging may determine similar albeit smaller relationships. Future studies which will incorporate control behavioral data will be needed to allow interaction modeling to assess disease specificity. Finally, the present relationships between CCD in lobule VI and outcome after stroke could be noncausal associations. We cannot infer causality. Noninvasive brain stimulation, either to upregulate/enhance cerebellar brain activity (Wessel & Hummel, 2018) or to perturb ongoing activity via virtual lesioning (Ziemann, 2010), might be one option to assess causality. Another option to explore potential noncausal effects might be mediation analyses which could answer to what extent the structural or functional state of reciprocal cortico‐ and subcortico‐cerebellar pathways (Bernard et al., 2013; Doyon et al., 2003; Guder et al., 2020; Schulz et al., 2015) could mediate the influence of regional CCD onto outcome after stroke.

5. CONCLUSION

In conclusion, we investigated lobule‐specific contralesional atrophy (as a surrogate of structural CCD) in a group of chronic stroke patients and healthy controls. Group comparisons revealed structural CCD of the whole cerebellum and particularly of lobules V and VI in stroke patients. The observed proportional volume reduction of lobule VI was associated with higher levels of disability and motor deficits. Lobule V and the whole cerebellum did not show similarly consistent diaschisis‐outcome relationships. Our results add to the understanding of cerebellar plasticity after stroke and might indicate that lobule VI is both prone to stroke‐related structural CCD and potentially involved in recovery processes and residual motor functioning after non‐cerebellar stroke.

AUTHOR CONTRIBUTIONS

Stephanie Guder made substantial contributions to the design, analysis and interpretation of data and drafting the article as well as revising it critically for important intellectual content. Fatemeh Sadeghi made substantial contributions to the analysis and interpretation of data and revising the article critically for important intellectual content. Simone Zittel, Fanny Quandt, Chi‐un Choe, Götz Thomalla, and Christian Gerloff made substantial contributions to the interpretation of the data and revising the article critically for important intellectual content. Marlene Bönstrup and Bastian Cheng made substantial contributions to the acquisition and interpretation of the data and revising the article critically for important intellectual content. Robert Schulz made substantial contributions to the design, acquisition, analysis and interpretation of data and drafting the article as well as revising it critically for important intellectual content. All authors approved the version to be published.

FUNDING INFORMATION

This work was supported by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation—178316478‐SFB936, projects C1 to C.G., C2 to G.T.); the Else Kröner‐Fresenius‐Stiftung (2020_EKES.16 to R.S., 2018_EKES.04 to C.U.C.); the European Union (EU‐Project euSNN, MSCA‐ITN‐ETN H2020‐860563 to S.Z. and C.G.); and the Gemeinnützige Hertie‐Stiftung (Hertie Network of Excellence in Clinical Neuroscience, to F.Q.).

CONFLICT OF INTEREST STATEMENT

The authors declare no conflict of interest.

Supporting information

DATA S1. Supporting Information.

Guder, S. , Sadeghi, F. , Zittel, S. , Quandt, F. , Choe, C. , Bönstrup, M. , Cheng, B. , Thomalla, G. , Gerloff, C. , & Schulz, R. (2023). Disability and persistent motor deficits are linked to structural crossed cerebellar diaschisis in chronic stroke. Human Brain Mapping, 44(16), 5336–5345. 10.1002/hbm.26434

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available from the corresponding author upon reasonable request.