Correlations between MRI and Information Processing Speed in MS: A Meta-Analysis

Table 1

Holm-Bonferroni Investigation into the relationships between whole-brain MRI measures and information processing tests.

MRI measure	Cognitive measure	Number of tests	Population	Number of null hypotheses rejected	Smallest 𝑃 value	Threshold	Number of NS results (𝑃>0.05)
T2LV	SDMT	18	All	8	<0.0001	0.0028	7
T2LV	SDMT	7	RRMS only	3	<0.0001	0.0073	2
T2LV	SDMT	9	Mixed only	6	<0.001	0.0057	3
T2LV	PASAT	27	All	4	<0.001	0.0019	15
T2LV	PASAT	13	RRMS only	0	<0.01	0.0039	8
T2LV	PASAT	12	Mixed only	4	<0.001	0.0043	6
Atrophy	SDMT	20	All	6	<0.001	0.0026	8
Atrophy	SDMT	7	RRMS only	1	<0.01	0.0073	5
Atrophy	SDMT	11	Mixed only	9	<0.001	0.0047	2
Atrophy	PASAT	23	All	4	<0.0001	0.0022	13
Atrophy	PASAT	10	RRMS only	0	0.045	0.0051	7
Atrophy	PASAT	11	Mixed only	4	<0.0001	0.0047	6

The most significant 𝑃 value in an analysis had to be lower than the threshold in order to reject any null hypotheses. Reported 𝑃 values are assumed equal to the maximum possible, for example, 𝑃<0.01 is tested as 𝑃=0.01. Where multiple 𝑃 values are reported for the same relationship (possibly adjusted versus unadjusted), the most insignificant 𝑃 value was used. Number of studies with RRMS + Mixed only do not necessarily sum to total studies, as some studies had a 100% SPMS or benign MS population. “Number of tests” strongly corresponds to number of studies; rarely, studies had data on subgroups that could not be combined.
MRI: magnetic resonance imaging; PASAT: paced auditory serial addition test; RRMS: relapse-remitting multiple sclerosis; SDMT: symbol digit modalities test; T2LV: T2-weighted lesion volume; NS: not significant.