Cerebro-cerebellar connectivity is increased in primary lateral sclerosis

Abstract

Increased functional connectivity in resting state networks was found in several studies of patients with motor neuron disorders, although diffusion tensor imaging studies consistently show loss of white matter integrity. To understand the relationship between structural connectivity and functional connectivity, we examined the structural connections between regions with altered functional connectivity in patients with primary lateral sclerosis (PLS), a long-lived motor neuron disease. Connectivity matrices were constructed from resting state fMRI in 16 PLS patients to identify areas of differing connectivity between patients and healthy controls. Probabilistic fiber tracking was used to examine structural connections between regions of differing connectivity. PLS patients had 12 regions with increased functional connectivity compared to controls, with a predominance of cerebro-cerebellar connections. Increased functional connectivity was strongest between the cerebellum and cortical motor areas and between the cerebellum and frontal and temporal cortex. Fiber tracking detected no difference in connections between regions with increased functional connectivity. We conclude that functional connectivity changes are not strongly based in structural connectivity. Increased functional connectivity may be caused by common inputs, or by reduced selectivity of cortical activation, which could result from loss of intracortical inhibition when cortical afferents are intact.

Keywords: AFNI, analysis of functional neuroimages; ALS, amyotrophic lateral sclerosis; ALSFRS-R, amyotrophic lateral sclerosis rating scale; ANCOVA, analysis of covariance; BOLD, blood oxygen-level dependent; Cerebellum; Connectivity; DTI, diffusion tensor imaging; Epi, echo planar imaging; FA, fractional anisotropy; FSL, FMRIB Software Library; FWE, family-wise error; MNI, Montreal Neurological Institute; Motor neuron disease; PLS, primary lateral sclerosis; Primary lateral sclerosis; ROI, region of interest; Resting state functional MRI; TBSS, tract based spatial statistics; TFCE, threshold-free cluster enhancement; TORTOISE, tolerably obsessive registration and tensor optimization indolent software ensemble; fMRI, functional magnetic resonance imaging.

Fig. S1

Correlation of increased functional connectivity between the 12 ROIs in PLS patients with A) disease duration and B) progression rate. Scale bar indicates Spearman correlation coefficient with warmer colors for more strongly and positively correlated regions. C) Red ROI pairs are those in which the significance of the correlation of increased functional connectivity with disease duration was <0.05 (uncorrected). Increased connectivity of cortical motor regions (cluster 3) with the right mid-temporal region was correlated with increased disease duration. D) The red ROI pair between two temporal regions is the only one in which the significance of the correlation of increased functional connectivity with progression was <0.05 (uncorrected).

Fig. 1

Regions with increased connectivity in PLS patients compared to healthy controls. A) “Seed” regions used to find connected ROIs with increased connectivity. Seeds are shown on axial (left) and sagittal (right) projections. Numbering corresponds to ROI position in the correlation matrix: 11. Right cerebellar seed, 12. Left cerebellar seed, 8. Cingulate seed,10. Right middle inferior temporal seed. B) Twelve ROIs, including original 4 seeds, with increased connectivity in PLS patients projected on axial sections. Sections shown in radiological convention with the right brain on the left side.

Fig. 2

K-means cluster analysis of functionally connected regions of interest. A) Elbow plot, showing 3 clusters is the best trade-off of variance explained to model complexity. B) Principle components analysis of the 12 × 12 ROI–ROI correlation matrix (averaged across groups); cluster membership is depicted simultaneously using color (green, red, blue). Cluster labeled #3, shown in green, consisted left precentral/SMA, right precentral, and right SMA ROIs; cluster in red labeled #2 consisted of right cingulate, right mid-frontal, right superior temporal, right mid-temporal, right putamen ROIs; cluster in blue labeled #1 consisted of left cerebellum, right cerebellum, right mid-temporal ROIs (SMA = supplementary motor area).

Fig. 3

Correlation matrices of the BOLD signal between 12 ROIs showing increased functional connectivity in A) healthy controls and B) PLS patients. Scale bar indicates correlation with warmer colors for more strongly correlated regions. C) Matrix showing ROIs in which the correlations differ between PLS patients and controls. Warmer colors indicate stronger correlations in PLS patients. The scale bar indicates T values. D) Matrix showing the correlation between the ALSFRS-R and connectivity of ROIs with the right cerebellar seed. Scale bar indicates the partial correlation coefficient. The arrow points to the column showing the correlation between the ALSFRS-R and connectivity of ROIs to the right cerebellar seed.

Fig. 4

Brain regions showing correlation between ALSFRS-R score and connectivity with right cerebellar seed. A) Right cerebellar seed, B) regions with significant correlation of ALSFRS-R with cerebellar connectivity. Colors indicate P value. Light blue = P < 0.05, dark blue = P < 0.01, purple = P < 0.005. Sagittal slices shown from right (top row) to left (bottom row).

Fig. 5

White matter regions with reduced fractional anisotropy in PLS patients compared to controls are shown in red. TBSS analysis (P < 0.05, TFCE with FWE correction for multiple connection). Axial sections are shown in radiological convention with right on the left side. MCP, middle cerebellar peduncle; SCP, superior cerebellar peduncle; RCP, right cerebral peduncle; TP, transverse pontine fibers; PLIC, posterior limb of the internal capsule; SCM, subcortical white matter. Arrowheads point to the corpus callosum.

Fig. 6

Results of probabilistic tractography using each of the 12 ROIs as a seed, with results expressed in terms of the probability of a connection. A) Median connection probability for each ROI × ROI combination across control subjects, and B) for PLS patients. Group differences are shown in C) as the PLS median−control median. No group differences survived correction for multiple comparisons, and the pattern of the results appeared unrelated to the group differences in functional connectivity (compared to Fig. 2).