Functional Connectivity Reveals Contralateral Cerebrocerebellar Circuits
Cerebrocerebellar connectivity was first explored in healthy control subjects. Hand motor seed regions in both hemispheres were defined from an independent fMRI study (see Methods and ). Subtraction between the two connectivity maps revealed the expected contralateral cerebrocerebellar connectivity (). Seed regions in the primary hand motor cortex correlated with the contralateral anterior lobe in lobules IV–VI in the cerebellum and with VIIIB in the posterior lobe, consistent with previous findings (Krienen and Buckner, 2009
Lesions to the Basilar Pons Disrupt Functional Connectivity
Intrinsic functional connectivity was examined in patients with lateralized pontine lesions (lesion locations illustrated in and ). Based on the anatomical characteristics of the cerebro-ponto-cerebellar circuit, we hypothesized that lateralized, focal lesions to the basilar pons would disrupt the functional connectivity between the ipsilesional motor cortex and the contralesional cerebellum, but spare the connectivity of the contralateral pathway. For initial analyses, the multiple MRI sessions were averaged within subject to boost the stability of the estimate at the expense of capturing variance due to recovery process, which is addressed later. Functional connectivity maps were first generated based on seed regions in the left and right motor cortex. Patients with left and right pontine lesions were visualized separately.
The correlation between left motor cortex and right cerebellum (LR) was reduced by left pontine lesions, while correlation between the right motor cortex and left cerebellum (RL) remained strong. This pattern was reversed in the patients with right pontine lesions. The cerebrocerebellar functional connectivity maps from several individual subjects clearly showed this pattern of disruption (), suggesting that intrinsic functional connectivity in this polysynaptic network is sensitive to the integrity of its specific anatomical pathway. Cerebrocerebellar connectivity maps were then averaged across subjects for the left lesion group, right lesion group, and healthy control group (). The average maps indicated that cerebrocerebellar functional connectivity was reduced in the lesioned circuit but remained strong in the unaffected circuit, for both left and right lesion groups.
Figure 4 Quantitative analysis of functional connectivity disruption in patients. (A) Cerebellar functional correlation to hand motor regions in the right lesion group, the healthy control group, and the left lesion group show reduced connectivity in the lesioned (more ...)
The cerebrocerebellar connectivity was then quantified for each subject. In control subjects, the cerebrocerebellar connectivity of the two circuits (LR and RL) was roughly symmetric at the group level (, middle plot, t = −1.48, p=0.17). In patients with lateralized pontine lesions, the predicted asymmetric cerebrocerebellar connectivity was observed in all but one of the individual subjects. The cerebrocerebellar connectivity pattern was reversed between the left and the right lesion subjects ().
To statistically test the effect, we combined all of the patients into a single group that had two measured values -- one for the lesioned circuit (lesioned) and one for the unaffected circuit (unaffected). Note that for some of these patients the lesioned circuit involved the right cerebellum and in others the left cerebellum. To accommodate this diversity, we yoked the 11 healthy control subjects to the patients, meaning that for each patient,, control measures for the lesioned and unaffected circuits were obtained for the same hemispheres as in the healthy control subjects. This allowed us to test the interaction between circuit (lesioned vs. unaffected) and group (patient vs. healthy control) while holding lesion laterality constant between groups.
The interaction effect was significant (F[1,20] = 9.2, p<0.01). Decomposing the effect, there was a main effect of circuit (lesioned vs. unaffected) for the patients (t = −3.3, p < 0.01 ) but not for the controls (t = 0.58, p = 0.57). Moreover, connectivity of the lesioned circuit in the patients was significantly reduced compared to the matched circuit in the healthy controls (unpaired t  = 2.4, p < 0.05), while the unaffected circuit in the patients showed no difference to the matched circuit in healthy controls (unpaired t = −0.34, p =0.73).
Patients were scanned five times following their stroke during a period of six months. For each of the five separate sessions, we compared the cerebrocerebellar connectivity of the unaffected circuit to the lesioned circuit (). We found that the connectivity strength in the unaffected circuit was stronger than the lesioned circuit across all time points. The effect was significant or showed a trend in four of the five time points (p=0.04, 0.08, 0.01, 0.01, 0.21 for the five time points, respectively). This analysis illustrates that the observations do not depend on averaging the multiple MRI sessions.
Figure 5 Cerebrocerebellar connectivity of the unaffected circuit (blue) is compared with the lesioned circuit (red) for each scanning session. The scanning sessions are labeled by the time following the stroke (x-axis label). Functional connectivity strength (more ...)
As a final analysis, functional connectivity between the left and right cerebral motor seed regions (MM) was also assessed. We found the M1-to-M1 correlation showed no difference between the patient and the control group (t =0.13, p=0.90), indicating that lateralized pontine lesion did not affect the hemispheric correlation in primary motor cortex although the cerebrocerebellar connectivity was disrupted.
Lesions to the Basilar Pons May Disrupt Non-motor Pathways
Our initial analyses focused on cerebrocerebellar motor circuits. However, the cerebral cortex forms circuits with the cerebellum that include non-motor pathways. Recent fcMRI studies have revealed specific cerebrocerebellar circuits that involve prefrontal regions and the contralateral cerebellum near Crus I and Crus II (Habas et al., 2009
; Krienen and Buckner, 2009
; O’Reilly et al., 2010
; Buckner et al., in press
). These circuits in the human are predicted by transneuronal tracing techniques in the monkey (Middleton and Strick 2001
) and involve paramedian and peripeduncular nuclei that are near the midline of the pons (Schmahmann and Pandya, 1997
). The patient lesions would thus be expected to also affect these circuits.
For this reason, we conducted a post-hoc
test to determine if lateralized pontine lesions also disrupt the functional connectivity between dorsolateral prefrontal cortex (DLPFC) and the cerebellum. Bilateral 8-mm radius, spherical DLPFC seed regions (MNI coordinates: ±42, 16, 36; coordinates reflect the centers of the regions) were defined based on a previous study (Krienen and Buckner, 2009
). Functional connectivity maps of these two seed regions were then subtracted.
Similar to the motor pathway, we found that correlation between the left DLPFC and the right cerebellum was reduced by left pontine infarction, while correlation between the right DLPFC and left cerebellum remained strong. This pattern was reversed in the patients with right pontine lesions (). These results suggest that functional connectivity between the DLPFC and the cerebellum also depends upon the integrity of the cortico-ponto-cerebellar circuit.
Figure 6 Pontine lesions also disrupt prefrontal-cerebellar circuits. Cerebellar functional correlations contrasting the right and left DLPFC regions in the left lesion group (left panel) and the right lesion group (right panel) both show reduced connectivity (more ...)