Crossed cerebellar diaschisis in patients with acute middle cerebral artery infarction: Occurrence and perfusion characteristics

Abstract

We aimed to investigate the overall prevalence and possible factors influencing the occurrence of crossed cerebellar diaschisis after acute middle cerebral artery infarction using whole-brain CT perfusion. A total of 156 patients with unilateral hypoperfusion of the middle cerebral artery territory formed the study cohort; 352 patients without hypoperfusion served as controls. We performed blinded reading of different perfusion maps for the presence of crossed cerebellar diaschisis and determined the relative supratentorial and cerebellar perfusion reduction. Moreover, imaging patterns (location and volume of hypoperfusion) and clinical factors (age, sex, time from symptom onset) resulting in crossed cerebellar diaschisis were analysed. Crossed cerebellar diaschisis was detected in 35.3% of the patients with middle cerebral artery infarction. Crossed cerebellar diaschisis was significantly associated with hypoperfusion involving the left hemisphere, the frontal lobe and the thalamus. The degree of the relative supratentorial perfusion reduction was significantly more pronounced in crossed cerebellar diaschisis-positive patients but did not correlate with the relative cerebellar perfusion reduction. Our data suggest that (i) crossed cerebellar diaschisis is a common feature after middle cerebral artery infarction which can robustly be detected using whole-brain CT perfusion, (ii) its occurrence is influenced by location and degree of the supratentorial perfusion reduction rather than infarct volume (iii) other clinical factors (age, sex and time from symptom onset) did not affect the occurrence of crossed cerebellar diaschisis.

Introduction

The first characterisation of ‘diaschisis’ traced back to von Monakow who described it in the 1870s as an ‘abolition of excitability’ and ‘functional standstill’ in an intact brain area distant from but linked to an area of cerebral damage. Baron et al.¹ transferred this concept in the 1980s to a phenomenon that was termed crossed cerebellar diaschisis (CCD). CCD refers to the association between a local supratentorial brain lesion and a simultaneous decrease of contralateral cerebellar blood flow and metabolic activity.² Current scientific concepts imply that interruption of corticopontocerebellar tracts causes a remote functional deactivation by a reduced excitatory input and a decreased cerebellar flow.³ CCD was reported in chronic diseases, such as brain tumors⁴ but also as a result of acute circuit inactivation, for example after stroke^5,6 and cerebral hemorrhage.⁷ The original definition of diaschisis was based on the notion that it was a transient and totally reversible event,⁸ whereas subsequent publications showed that it can persist for years and lead to irrevocable degeneration.^9–11 In the first hours after stroke, however, it has been shown that diaschisis is potentially reversible if supratentorial reperfusion can be achieved.^5,12 The pathophysiological concept of diaschisis is nowadays widely accepted and many articles concerning different aspects of this phenomenon have been published meanwhile. Controversies prevail, though, e.g. in regard to the frequency of CCD following different forms of supratentorial damage, its prognostic value, its dependency on the location and size of the supratentorial lesion and its clinical correlates.^6,13,14

Imaging studies on CCD have mainly involved nuclear medicine-based techniques like SPECT, PET and Xenon-computed tomography. All of them are used as reference methods for perfusion imaging but are not available in routine clinical care. Therefore, only small patient collectives have been evaluated with regard to CCD after acute supratentorial damage by these methods so far. Advances in imaging technology now allow whole-brain CT perfusion (WB-CTP),¹⁵ a method, which is becoming increasingly available and is part of the acute stroke workup in many centers. Using this methodology, it has been possible to robustly study the cerebellar perfusion¹⁶ and to detect CCD after acute supratentorial damage.^7,17

In our study, we screened a large patient cohort and aimed to (i) investigate the overall prevalence of CCD after acute infarction in the middle cerebral artery (MCA) territory using WB-CTP and to (ii) determine potential factors influencing the occurrence of CCD.

Materials and methods

Study population and patient selection

From March 2009 to January 2014, 1644 consecutive patients who were admitted to our institution for suspected stroke and who underwent emergency multimodal stroke CT were evaluated. The study was designed as a retrospective single-center study at a large university hospital. The institutional review board of the Ludwig-Maximilians University Munich (Ethikkommission der Medizinischen Fakultät der Ludwig-Maximilians Universität München) approved the study according to the Helsinki Declaration of 1975 (and as revised in 2013) and waived requirement for informed consent. The final study population was selected according to the following criteria:

Inclusion criterion

Acute occlusion of the ICA, the carotid-T and/or the M1 and M2 segment of the MCA on CT angiography (CTA) with a concomitant perfusion deficit in the MCA territory on CTP

Exclusion criteria

1. Absence of confirmed infarction of the MCA territory by follow-up CT/MRI.

2. Any abnormality of the vertebrobasilar arteries as determined by CTA (e.g. hypoplasia, stenosis, occlusion).

3. Any pathology of the cerebellum in non-enhanced CT (NECT) of the brain (e.g. prior ischemia, hemorrhage, tumor).

4. Cerebellar ischemia on follow-up CT/MRI.

5. CTP coverage of less than 50% of the cerebellum or poor image quality.

The study was designed as a case–control study with a ratio of cases to controls of 1:2 in order to prevent biased reading.¹⁸ WB-CTP datasets of the control group had no cerebrovascular abnormalities on WB-CTP as well as on follow-up MRI.

CT examination protocol

The CT imaging protocol consisted of a NECT to exclude intracerebral hemorrhage, a supra-aortic CTA and a WB-CTP. The examinations were performed on the following multislice CT scanners: SOMATOM Definition AS+, a 128 slice CT scanner; SOMATOM Definition Flash, a 128 slice dual source CT scanner; SOMATOM Definition Edge, a 128 slice CT scanner (all Siemens Healthcare, Erlangen, Germany). WB-CTP was acquired with 0.6 mm collimation and scan coverage of 100 mm in the z-axis by means of a toggling table technique. One scan was acquired every 1.5 s with the following imaging parameters: tube voltage 80 kV; tube current 200 mAs; and CTDI_vol 276.21 mGy. Thirty-five millilitres of highly iodinated contrast agent were administered intravenously at a flow rate of 5 ml/s, followed by a saline flush of 40 ml at 5 ml/s; Axial slices (31) with a thickness of 10 mm and an increment of 3 mm were reconstructed from the dataset.

CT perfusion image processing

The processing of source images was performed with the SYNGO Volume Perfusion CT Neuro software using a semi-automated deconvolution algorithm (Auto Stroke MTT) on a dedicated workstation (Syngo MMWP, VA 21A; Siemens Healthcare, Erlangen, Germany). The arterial input function was selected as described earlier.¹⁹ Briefly, a vendor given algorithm was used to select a region of interest (ROI) within the earliest-appearing arteries at a representative slice at the level of the basal ganglia. The venous output region was selected from the superior sagittal sinus. A series of 31 color-coded slices was generated for each of the hemodynamic parameters cerebral blood flow (CBF), cerebral blood volume (CBV), mean transit time (MTT), time to drain (TTD) and time to peak (TTP). CBV is the whole amount of blood in a defined unit of tissue. In contrast to CBV, CBF, MTT, TTP and TTD are related to the passage of blood in a given time window. Due to changes in the software features during the period of data acquisition (2009–2014), MTT, TTD and TTP were not available for all but in 99.0%, 87.7% and 24.0% of the patients, respectively. CBF and CBV were assessed for all datasets. The color slices were saved in DICOM format and displayed on default window settings for further evaluation.

Follow-up imaging

For patients with MCA infarction, follow-up imaging consisted of MRI in 88 (56.4%) patients and NECT in 68 (43.6%) patients, respectively. The MRI follow-up examination comprised a T2* sequence, a fluid-attenuated inversion recovery (FLAIR) sequence, a diffusion-weighted imaging (DWI) sequence and a time-of-flight (TOF)-MR angiography. The median time delay between CT perfusion scan and follow-up imaging was two days (range: 1–49 days) for MRI and one day (range: 1–16 days) for NECT.

Image analysis

Initially, a study population was created by random selection of patients (N_p = 156) and controls with a ratio of 1:2. Subsequently, the respective CTP data sets (CBF, CBV, MTT, TTD and TTP) were converted into JPEG files and inserted into PowerPoint (Microsoft PowerPoint 2013, Microsoft Co., Redmond/US-WA) for further processing. The JPEG files were saved in the default format without the possibility of changing the window setting. All images were presented to the readers after having been cropped to an extent that only the cerebellar, but not the supratentorial brain parenchyma was visible (performed by CB). The assessment of presence/absence of CCD was subsequently performed qualitatively by two independent raters (one neurologist with eight years (LvB) and one radiologist with seven years (WS) of experience in CTP reading, respectively) who were blinded to the clinical data. In case of disagreement, a consensus was reached in a separate session.

In the following analysis, only perfusion anomalies in the cerebellar hemisphere contralateral to the supratentorial lesion were counted as CCD positive. CCD was rated positive if at least two of the perfusion maps (irrespective of which parameters) showed a deficit on the same side. This approach has already been applied by Thierfelder et al.^16,20 with the objective of achieving valid results for both sensitivity and specificity.

The degree and extent of the supratentorial ischemic lesion on admission were quantified on each of the existing CTP maps by applying the Alberta stroke program early computed tomography score for CT perfusion (CTP-ASPECTS), a topographical scoring system, which provides localisation-weighted values ranging from 0 to 10.²¹ Additionally, we extended the ASPECT score by adding the thalamus as a further anatomical region of interest, as a perfusion deficit in this location was described to be relevant for the induction of cerebellar diachisis.^22,23

Final infarct volume was determined either on follow-up MR or on follow-up NECT using a volumetric approach as previously described.¹⁹ Briefly, regions of interest around the infarct lesions were delineated on all relevant slices using OsiriX (OsiriX V.4.0 imaging software 2011) on a 27-in. iMac computer (Apple Inc., Cuperino, CA, USA). Infarct volume was calculated in cm³/ml by multiplying the areas by the slice thickness.

The software Syngo.via Single Sign On (Siemens Healthcare, Erlangen, Germany) was used for the calculation of the absolute perfusion values based on the raw data which were still available for 116 of 156 patients (74.4%). We chose a semi-quantitative approach, which was described previously.¹⁷ Briefly, the area of the abnormal perfusion was segmented manually in the affected hemisphere, and subsequently was mirrored to the contralateral hemisphere. A representative slice, showing the greatest interhemispheric difference was chosen and differences in the absolute perfusion parameters were determined. The same procedure was adopted for the cerebellar hemispheres. Here, circular ROIs were used to obtain quantitative perfusion values.

Perfusion asymmetry was assessed as follows:

Cerebrum: TTD_ipsilateral−TTD_{contralateral}

         MTT_ipsilateral−MTT_{contralateral}

      (CBF_{contralateral}−CBF_ipsilateral)/

         CBF_{contralateral} × 100% (CBF reduction rate)

      (CBV_{contralateral}−CBV_ipsilateral)/

         CBV_{contralateral} × 100% (CBV reduction rate)

Cerebellum: TTD_{contralateral}−TTD_ipsilateral

       MTT_{contralateral}−MTT_ipsilateral

  (CBF_ipsilateral−CBF_{contralateral})/    CBF_ipsilateral × 100% (CBF reduction rate)

  (CBV_ipsilateral−CBV_{contralateral})/    CBV_ipsilateral × 100% (CBV reduction rate)

Statistical analysis

All statistical analyses were performed using Excel (Microsoft Excel 2013, Microsoft Co., Redmond/US-WA) and MedCalc (MedCalc Software, Ostend/Belgium). On the basis of the two independent readings, the degree of inter-rater agreement in the evaluation of perfusion maps was assessed by calculation of the Kappa coefficient. In addition, the proportion of patients with MCA infarct diagnosed as CCD positive was determined in general and for each of the perfusion parameters. To test for normal distribution the Shapiro-Wilk test was performed. Given a non-normal distribution, we applied the Mann-Whitney-U test to identify significant differences between patients classified as CCD positive and CCD negative classified patients. The association between CCD and the presence of ischemic lesions on CTP in the predefined anatomical cerebral regions was elucidated by use of the Chi-squared test and Fisher’s exact test. The Spearman’s rank correlation was used to analyse the relationship between supra- and infratentorial perfusion values. The significance level for all tests was set at P < 0.05. In case of multiple testing, the significance level was adjusted according to Bonferroni. Data are expressed as mean ± standard deviation/95% confidence interval or as median ± interquartile range depending on the distribution of data.

Results

Patient characteristics

The final patient cohort consisted of 156 individuals who were included in the further analysis. The patient flow chart is illustrated in Figure 1. Mean age (±SD) was 69.7 years (±14.9) and 63.3 years (±15.8) in the patient group and in the control group, respectively. The patient group included 69 male subjects (44%) compared with 87 in the control group (56%). Time from symptom onset did not differ significantly between the cohorts (228 ± 213 min vs. 218 ± 323 min, P = 0.7471. In the patient group, MCA infarcts were located in the left cerebral hemisphere in 52.6% of cases. Mean final infarct volume (±SD) was 56.5 ml (±86.6) and median ASPECTS score was 3 (interquartile range: 1–6). The most common diagnoses in the control group included transient ischemic attacks (44.5%), epileptic seizures (16.1%), encephalitis/myelitis/meningitis (5.8%), encephalopathy (5.2%) and cerebral hemorrhage (3.9%). Patient characteristics are summarised in Table 1.

Figure 1.

Patient selection.

Table 1.

Characteristics of the study population.

	Patient group		Control group
	Value	Range	Value	Range
N	156		312
Age (years) ± SD	69.7 ± 14.9	28–97	63.3 ± 15.8	27–110
Male gender (%)	44	–	56	–
Time from symptom onset (min) ± SD	228 ± 213a	34–1356	218 ± 323b	53–2462
Infarction on left hemisphere (%)	52.6	–	–	–
ASPECTS (median) ± interquartile range	3 ± 1/6	0–10	–	–
Final infarction volume (ml) ± SD	74.9 ± 100.4	0.1–496.1	–	–

SD: standard deviation; ASPECTS: Alberta stroke program early computed tomography score.

^aAvailable for 60.3% of subjects.

^bAvailable for 41.0% of subjects.

Inter-rater agreement

We assessed the inter-rater agreement in the detection of CCD. The kappa statistic calculated by means of the two independent readings yielded a value of 0.73 which can be interpreted as a good degree of agreement.²⁴

Frequency of CCD

The number of CCD-positive and CCD-negative CTP maps was determined on the basis of the consensus reading. Results are illustrated in Table 2. Within the patient group, 55/156 (35.3%) subjects were rated CCD positive, while 101/156 (64.7%) subjects were diagnosed as CCD negative. Among controls, 39 CTP maps (12.5%) were rated false-positive for CCD, while the remaining 273 CTP maps (87.5%) were correctly identified as CCD negative. CCD was detected with a significantly higher frequency in the presence of MCA infarction than in controls (35.5% vs. 12.5%, P < 0.0001).

Table 2.

Proportion of CCD+ and CCD− cases among the patient and control group.

	CCD+	CCD−	P
Patient group (N = 156)	55 (35.3%)	101 (64.7%)	<0.0001a
Control group (N = 312)	39 (12.5%)	273 (87.5%)

CCD: crossed cerebellar diaschisis.

^aStatistically significant (Chi-squared test).

The CTP map with the highest detection rate of CCD among the subjects with MCA infarct was CBF (35.0%), followed by TTD (31.4%), CBV (26.3%), MTT (22.4%) and TTP (17.5%). For the differentiation between CCD positive and negative cases, TTP showed a high diagnostic value of 98.6%. Similarly, CBV performed considerably well with 94.2%, while the other parameters CBF (87.3%), TTD (85.5%) and MTT (85.5%) ranged at a lower level. Figure 2 shows three representative examples of CCD-positive examinations.

Figure 2.

Representative cases of MCA infarction and CCD are shown in Figure 2 (a–c). CTP maps (CBF, CBV, MTT and TTD, right panel: higher resolution) of the cerebellar perfusion on admission are shown in the red frame. The infarcted tissue (CBF upon admission, corresponding follow-up MRI-DWI) is demonstrated within the grey frame. (a) Ninety year-old woman with acute right-sided hemiparesis, left-sided gaze preference and expressive aphasia. Time from symptom onset to CTP imaging was 75 min and to MRI follow-up 67 h, respectively. (b) Sixty-nine year-old woman with acute right-sided hemiparesis and global aphasia. Time from symptom onset to CTP imaging was 150 min and to follow-up MRI 71 h, respectively. (c) Sixty-five year-old woman with acute left-sided hemiplegia, dysarthria and a neglect. Time from symptom onset to CTP imaging was 120 min and to follow-up MRI 100 h. In all patients, rtPA was administered immediately after CT workup.

Determining factors for the occurrence of CCD

Among patients with MCA infarction, we analysed factors with potential influence on the occurrence of CCD. Results are presented in Table 3. A left-sided MCA infarction was significantly associated with the occurrence of CCD (CCD positive: 56.4% left hemisphere vs. 43.6% right hemisphere; CCD negative: 50.5% left hemisphere vs. 49.5% right hemisphere, P = 0.0427). A higher mean ASPECTS score was found for CCD-positive patients (2.6 (2.0;3.1) for CCD positive vs. 3.5 (3.0;4.0) for CCD negative, P = 0.0283). Other variables (‘time from symptom onset to CT examination’, ‘age’, ‘gender’ and ‘volume of supratentorial infarction’) did not reach the predefined level of significance of P < 0.05. After the Bonferroni correction for multiple testing, only a left-sided infarction but not the ASPECTS score remained statistically significant.

Table 3.

Comparison of clinical features and infarct variables between CCD+ and CCD− cases.

	CCD+ (N = 55)	CCD− (N = 101)	P
Volume of supratentorial infarction (ml)a	81.3 (55.6;107.0)	71.4 (51.0;91.9)	0.1297b
Supratentorial infarction of left hemisphere (%)	56.4	50.5	0.0427c,d
ASPECTSa	2.6 (2.0;3.1)	3.5 (3.0;4.0)	0.0283b,d
Time from symptom onset (min)a	205.5 (146.2;264.8)	239.7 (174.9;304.5)	0.5622b
Age (years)a	67.0 (63.7;70.4)	72.4 (68.5;76.2)	0.0815b
Male gender (%)	45.5	43.6	0.9534c

CCD: crossed cerebellar diaschisis; ASPECTS: Alberta stroke program early computed tomography score.

^aValues are mean (95% confidence interval).

^bMann-Whitney test.

^cChi-squared test.

^dStatistically significant.

Dichotomised univariate associations between infarct locations as assessed by ASPECTS and CCD were explored by applying the Chi-squared test to the data of MCA-infarct patients. Results are shown in Table 4. Cerebral hypoperfusion involving the following territories were significantly associated with CCD: M 1 (anterior MCA cortex; CCD-positive examinations: 94.5% vs. CCD-negative examinations: 80.2%, P = 0.0175), M 4 (anterior MCA territory superior to M 1) (CCD positive: 98.2% vs. CCD negative: 82.2%, P = 0.0036), the internal capsule (CCD positive: 67.3% vs. CCD negative: 44.6%, P = 0.0109) and the thalamus (CCD positive: 69.1% vs. CCD negative: 33.7%, P < 0.0001). No significant association was detected for other cortical regions (insula, M 2: cortex lateral to insular ribbon, M 3: posterior MCA cortex, M 5: lateral MCA territory superior to M 2, M 6: posterior MCA territory superior to M 3) and subcortical regions (caudate nucleus, lentiform nucleus). After a Bonferroni correction for multiple testing, only the thalamus and the ASPECTS-defined region M4 which is located in the frontal lobe remained statistically significant. Figures 2(c) and, in more detail, Figure 3 show examples of an ipsilateral thalamic hypoperfusion in a patient with CCD.

Table 4.

Influence of supratentorial infarct location on the incidence of CCD.

Anatomical structures involved in supratentorial infarction		CCD+ (N = 55)	CCD− (N = 101)	P
A S P E C T S	Caudate nucleus	63.6% (35/55)	51.5% (52/101)	0.1966a
	Lentiform nucleus	63.6% (35/55)	56.4% (57/101)	0.4819a
	Internal capsule	67.3% (37/55)	44.6% (45/101)	0.0109a
	Insula	94.6% (52/55)	90.1% (91/101)	0.5453b
	M 1	94.6% (52/55)	80.2% (81/101)	0.0175b
	M 2	98.2% (54/55)	92.1% (93/101)	0.1610b
	M 3	92.7% (51/55)	83.2% (84/101)	0.1396b
	M 4	98.2% (54/55)	82.2% (83/101)	0.0036b,c
	M 5	98.2% (54/55)	94.1% (95/101)	0.4225b
	M 6	98.2% (54/55)	89.1% (90/101)	0.0574b
Thalamus		69.1% (38/55)	33.7% (34/101)	<0.0001a,c

CCD: crossed cerebellar diaschisis; ASPECTS: Alberta stroke program early computed tomography score.

^aChi-squared test.

^bFisher’s exact test.

^cStatistically significant.

Figure 3.

A representative case of MCA infarction, ipsilateral thalamic hypoperfusion and CCD is shown in Figure 3. CTP maps (CBF, CBV, MTT and TTD) of the supratentorial perfusion including the thalamus as well as the respective follow-up MRI (DWI and FLAIR) are shown in the red frame. White arrows indicate the area of thalamic hypoperfusion. The cerebrellar hypoperfusion (left side: CBF, right side: higher resolution CBF, CBV, MTT and TTD) are shown in the gray frame. Images were obtained from a 48 year-old male with acute left-sided hemiparesis due to a right-sided M1-occlusion. Time from symptom onset to CTP imaging was 190 min and to MRI follow-up 64 h, respectively. After CT-workup, rtPA was administered immediately and subsequent (successful) mechanical recanalisation of the M1-segment was performed.

Quantitative analysis of perfusion values

Quantification of the relative supratentorial cerebral perfusion parameters revealed that that MTT and TTP prolongation as well as CBF reduction is more pronounced in CCD-positive patients than in CCD-negative patients (Table 5). Significant differences between CCD-positive and -negative patients could be detected for MTT (0.48 (0.09;0.75) for CCD positive vs. 0.10 (0.19;0.43) for CCD negative; P = 0.0005), TTD (0.62 (0.30;1.04) for CCD positive vs. 0.33 (−0.04;0.61) for CCD negative; P = 0.0017) and CBF (19.28 (15.78;26.69) for CCD positive vs. 14.15 (4.66;21.75) for CCD negative; P = 0.0032). CBV, however, showed no significant difference between these groups (11.67 (7.98;18.34) for CCD positive vs. 11.86 (3.68;20.31) for CCD negative; P = 0.9217).

Table 5.

Comparison of supratentorial perfusion reduction between CCD+ and CCD− cases.

Cerebral perfusion parameters	CCD+ (N = 43)	CCD−(N = 73)	P
	Median (95% CI)	Median (95% CI)
Δ MTT (s)	0.48 (0.09;0.76)	0.10 (−0.19;0.43)	0.0005a
Δ TTD (s)	0.62 (0.30;1.04)	0.33 (−0.04;0.61)	0.0017a
CBV reduction rate (%)	11.67 (7.98;18.34)	11.86 (3.68;20.31)	0.9217
CBF reduction rate (%)	19.28 (15.78;26.69)	14.15 (4.66;21.75)	0.0032a

MTT: mean transit time; TTD: time to drain; CBF: cerebral blood flow; CBV: cerebral blood volume.

Δ MTT: MTT_ipsilateral − MTT_{contralateral},

Δ TTD: TTD_ipsilateral − TTD_{contralateral},

CBV reduction rate: (CBV_{contralateral} − CBV_ipsilateral)/CBV_{contralateral} × 100%,

CBF reduction rate: (CBF_{contralateral} − CBF_ipsilateral)/CBF_{contralateral} × 100%,

^aStatistically significant.

Among CCD-positive cases, no significant correlation could be detected between the degree of the supratentorial perfusion and the cerebellar perfusion reduction (Table 6).

Table 6.

Correlation of cerebral and cerebellar perfusion parameters in CCD+ cases.

	Spearman’s Rho correlation coefficient (P)
	Cerebellar Δ MTT	Cerebellar Δ TTD	Cerebellar CBV reduction rate	Cerebellar CBF reduction rate
Cerebral Δ MTT	0.132 (0.399)	0.275 (0.074)	0.049 (0.758)	0.155 (0.321)
Cerebral Δ TTD	0.132 (0.398)	0.107 (0.494)	0.090 (0.565)	0.198 (0.204)
Cerebral CBV reduction rate	0.097 (0.538)	0.052 (0.742)	−0.027 (0.862)	0.051 (0.744)
Cerebral CBF reduction rate	0.302 (0.049)	0.370 (0.015)	0.021 (0.892)	0.232 (0.134)

MTT: mean transit time; TTD: time to drain; CBF: cerebral blood flow; CBV: cerebral blood volume.

Cerebral Δ MTT: MTT_ipsilateral − MTT_{contralateral}

Cerebral Δ TTD: TTD_ipsilateral − TTD_{contralateral},

Cerebral CBV reduction rate: (CBV_{contralateral} − CBV_ipsilateral)/CBV_{contralateral} × 100%,

Cerebral CBF reduction rate: (CBF_{contralateral} − CBF_ipsilateral)/CBF_{contralateral} × 100%,

Cerebellar Δ MTT: MTT_{contralateral} − MTT_ipsilateral,

Cerebellar Δ TTD: TTD_{contralateral} − TTD_ipsilateral,

Cerebellar CBV reduction rate: (CBV_ipsilateral − CBV_{contralateral})/CBV_ipsilateral × 100%,

Cerebellar CBF reduction rate: (CBF_ipsilateral − CBF_{contralateral})/CBF_ipsilateral × 100%.

Discussion

WB-CTP offers a robust method to screen for CCD in a large patient collective. In our cohort, perfusion abnormalities of the cerebellum consistent with CCD were observed in 35.3% of all patients. The perfusion parameters CBF and TTD reached the highest detection rates, whereas TTP and CBV best distinguished between CCD-positive and -negative cases. The location of the supratentorial perfusion deficit, rather than its size, as well as the severity of cerebral hypoperfusion was decisive for the occurrence of CCD. We could show that a location on the left side, in the frontal lobe and the thalamus is associated with the evolution of CCD.

The frequency of CCD after MCA infarction in our study lies within the published range of CCD after supratentorial strokes (15.61%–46.2%).^{13,14,22,25–27} The results of these studies, however, were predominantly obtained by analyses of small study populations (range: 18–113)^{13,14,17,22,25,26,28} and/or by inclusion of various vascular territories^{1,13,25,27–29} and by inclusion of acute and chronic infarcts.^26,27 The strokes we investigated, however, were acute strokes restricted to the MCA territory to avoid a dispersion of the results by inclusion of different infarct stages and vascular territories. By screening a large patient collective, we can thus further extend and substantiate the observation that CCD is a common phenomenon after MCA infarction and that WB-CTP provides an appropriate tool to investigate its hemodynamic impact on cerebellar perfusion after cerebral infarction.

In our study, we found a high rate of thalamic hypoperfusion in patients with MCA stroke. In principle, occlusions of the internal carotid artery could lead to an insufficient thalamic blood supply via the posterior communicating artery as well as via the anterior choroid artery. However, blood flow via the contralateral side should compensate for this deficit, and thalamic infarction usually relates to infarctions of the tuberothalamic artery (arising from the posterior communicating artery), the paramedian, the inferolateral artery or the posterior choridal artery (arising from the posterior cerebral artery).³⁰ In our study, ipsilateral thalamic perfusion alterations were also observed in patients without carotid occlusion (data not shown). Notably, in all cases with an ipsilateral thalamic perfusion alteration on the initial CTP, no thalamic infarction was detected in the follow-up imaging. According to the literature, not only cerebellar diachisis but also ipsilateral thalamic hypoperfusion have been described after cortical³¹ and capsular³² infarctions in rats and after MCA infarctions in humans.^33,34 Similarly, in status epilepticus, diffusion restrictions indicating CCD as well as an ipsilateral thalamic involvement have been described.^35,36 Furthermore, crossed cerebello-thalamo-cerebral diachisis as indicated by an FDG hypometabolism in the contralateral thalamus and cerebral cortex has been reported after acute cerebellar injury using FDG-PET.³⁷ Therefore, the observed perfusion alteration in the thalamus most likely relates to an ipsilateral thalamic diachisis.

According to current knowledge, the interruption of fiber tracts projecting via pontine nuclei to the contralateral cerebellar cortex and to the deep cerebellar nuclei lead to a reduced activation of the cortical Purkinje cells as well as a reduced activity of glutaminergic projection neurons from the deep cerebellar nuclei to the thalamus.^38,39 This results in a functional deactivation and ultimately to vasoconstriction which in turn results in a profound reduction of CBF and CBV.⁴⁰

The relatively high detection rate of CBF in comparison to the other perfusion parameters supports this theory. Due to the concomitant reduction of CBF and CBV, changes in time-based perfusion maps are not expected.⁴¹ However, we and others,^7,27 do indeed find alterations in time-based MRI²⁷ and CTP-based perfusion maps like MTT, TTP and TTD.^7,17 Importantly, we ruled out that the perfusion changes in the time-dependent maps are related to vascular disease or ischemia in the posterior circulation.

In the literature, disagreement exists whether infarction volume influences CCD occurrence and severity. Kim et al.¹³ reported that the location rather than the extent or severity of the lesion may be the major determinant for the occurrence and magnitude of CCD in patients with cerebral infarction. Sobesky et al.,⁵ however, found that acute but not chronic CCD was closely related to the volume of supratentorial hypoperfusion. Lin et al.²⁷ showed in a small study that infarct volume is related to the development of CCD. In our study on acute MCA stroke, however, we could find no association between infarct size and occurrence of CCD. In this study, we could demonstrate that a more pronounced supratentorial perfusion alteration seen in MTT, TTD or CBF is positively correlated with the evolution of CCD, which is in line with a previous SPECT study demonstrating that the degree of hypoperfusion is correlated with the occurrence of CCD.²⁵ However, we did not find a correlation between the degree of the supratentorial perfusion alteration and the degree of the cerebellar hypoperfusion.

Cerebral hypoperfusion in the frontal lobe and the thalamus were associated with CCD which is in agreement with previously published results of smaller collectives.^{10,13,22,29,42,43} Pantano et al.⁴⁴ demonstrated that CCD was more prominent when the supratentorial infarct involved the internal capsule or the cortical mantle extensively. They also suggested that destruction of the pyramidal tract is neither necessary nor sufficient to induce CCD. Gold and Lauritzen³ showed that decreases of activity involving the frontal cortex produced the largest decrease in contralateral cerebellar electrical activity and blood flow. It has been shown that small infarcts, if located in a strategically relevant thalamic area (e.g. nucleus ventralis intermedius), are sufficient to elicit a relevant decrease of the contralateral cerebellar blood flow and metabolism.^22,23 Overall, these data are in line with our observation that there is an association between anatomic region and the incidence of CCD which is not the case for the size of the cerebral hypoperfusion.

The structural basis accounting for the development of CCD by cerebral hypoperfusion involving the frontal lobe and the thalamus might be the dentatorubrothalamic pathway, specific thalamic projections and the corticopontocerebellar tract.^23,45 Furthermore, it is known that the cerebellar-cortical pathways are structured in reciprocal organised loops and that the cerebellar hemispheres in particular are interconnected to the frontal cortex.⁴⁶ In our cohort, a significantly higher proportion of CCD occurred after left-sided MCA infarctions. This observation may indicate that frontothalamocerebellar circuits which are supposed to be intricately involved in language,⁴⁷ complex motor functions,⁴⁸ verbal working performance⁴⁹ account for the development of CCD. Based on our findings, it can be suggested that the amount of interconnection of frontal cortex and cerebellar hemisphere is more intense for the left frontal cortex.

Subacute neurologic function and recovery have been shown to be worse in patients with stroke and CCD compared with those without CCD.⁵⁰ Although no reports to date show that cerebellar hypoperfusion from CCD is associated with permanent infarction, chronic deafferention results in measurable structural abnormalities.⁵¹ A serial PET study with longitudinal follow-up at multiple time points after thrombolytic therapy also highlighted the usefulness of CCD as an indicator of clinical outcomes.⁵ Our data indicate that WB-CTP evaluation might be a potent tool to clarify the role of CCD for patient outcome after MCA infarction.

Our data must be interpreted in the context of the study design. This is a retrospective single-center study on the occurrence of CCD after MCA infarction which is based on a dichotomised univariate descriptive statistical analysis. Consequently, no exact predictions but only significant associations can be revealed by our investigations. We did not analyse the potential impact of CCD on patient’s clinic or outcome, although these are important aspects for future studies. However, our study demonstrates that such studies could well be performed using the WB-CTP. We blinded our readers to clinical data and presence of a supratentorial infarction to minimise bias. Moreover, we enrolled a large patient cohort which exceeded the size of study populations investigated in terms of CCD so far (mean: 51.1; range: 18–113).^{13,14,17,22,25,26,28,52} Still, we cannot fully rule out the possibility of having underestimated the frequency of CCD as we counted only those cases as CCD positive, that show hemodynamic alterations suggestive of CCD in at least two perfusion parameter maps. However, artifacts on single perfusion maps can often be observed in CT imaging of the posterior fossa and this approach helped to limit the number of false-positive findings.

Moreover, the sensitivity might have been negatively influenced by the fact that JPEGs of the raw perfusion maps were presented to the readers which could not be windowed for brightness and contrast. The JPEG files, however, allowed us to easily modify the images to blind the readers to the supratentorial brain sections.

CTP alterations suggestive of CCD were also observed in subjects of the control group. These might be related to artifacts. Moreover, transient ischemic attacks or epileptic seizures, two of the two main reasons for admission within the control group, could potentially result in CCD and cause a substantial rate of false-positive results.

Conclusion

CCD is a common phenomenon after MCA infarcts and leads to a cerebellar hypoperfusion which can be detected by decreased CBF, CBV and/or by prolonged TTP, TTD and MTT. This phenomenon needs to be considered when interpreting CTP in acute supratentorial stroke in order to correctly classify contralateral cerebellar perfusion deficits. Prospective studies with larger cohorts and profound clinical assessment are necessary in order to further determine the clinical relevance of the observed phenomenon. In particular, patient relevant aspects like the immediate and long-term clinical correlates of CCD and its potential influence on outcome need to be subject of further research. For this purpose, an analysis of the WB-CTP images performed for the initial stroke workup seems to be a robust way to screen for CCD.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Authors’ contributions

WHS and CB: contributed equally to this article. Conceived and designed the experiments: LVB, WHS. Performed the experiments: CB, AB, LVB, WHS. Analysed the data: CB, AB, WHS, LVB. Wrote (and revised) the manuscript: WHS, CB, LVB, AP, KMT, AS, MFR, JH, BEW.