Accumulating evidence from case-control and population studies suggests attention-deficit/hyperactivity disorder (ADHD) confers a 2- to 5-fold risk of all-cause dementia later in life. Here, we investigate vascular burden as a potential mediator of this relationship, because vascular integrity is well known to be compromised in ADHD (due to chronic obesity, diabetes, and hypertension) and is also a robust risk factor for neurodegeneration (due to reduced cerebral blood flow). We use brain white matter hyperintensities (WMH) as a measure of vascular burden.

Thirty-nine adults aged 48-81 years with clinical ADHD, and 37 matched controls, completed neuropsychological testing and 1.5 T structural neuroimaging. None had stroke. Cognitive tests were demographically-adjusted to Z scores using regression-based norms generated from the control group, and averaged across tests within domains of short- and long-term verbal memory (forward digit span, California Verbal Learning Test, Logical Memory), visual memory (Visual Recognition, Rey Complex Figure), processing speed (coding, trails A, Stroop word-reading and color-naming), language (Boston Naming Test, semantic fluency), visuoconstruction (clock drawing, Rey Complex Figure copy), and executive function (backward digit span, trails B, phonemic fluency, Stroop inhibition, Wisconsin Card Sorting Test). Total WMH volumes (i.e., combined periventricular and deep) within subcortical, temporal, frontal, parietal, and occipital regions were individually divided by regional volumes to produce a proportion of each region representing WMH, then log-transformed to correct for skew. Age-corrected linear regression quantified total effects of ADHD on cognition; when these were significant, mediation models quantified the direct effects of ADHD on WMH volumes and the direct effect of WMH volumes on cognition. Sobel’s test estimated indirect effects of ADHD on cognition via WMH.

Group had a significant total effect on Processing Speed (ß=-1.154, p<.001) and on Executive Functioning (ß=-0.587, p=.004), where ADHD participants had lower composite scores (M=-1.10, SD=1.76 and M=-0.54, SD=1.14 respectively) than controls (M=0.02, SD=0.74; M=0.00, SD=0.49). Only frontal-lobe WMH had direct effects on Processing Speed (ß=-0.315, p=.012) and Executive Functioning (ß=-0.273, p<.001). The direct effect of ADHD on frontal WMH was significant (ß=-0.734, p=.016), and Sobel’s tests supported an indirect effect of ADHD on Executive Functioning (z=2.079, p=.038) but not Processing Speed (z=1.785, p=.074) via WMH. Because the effect of ADHD on WMH was negative (i.e., fewer WMH in ADHD) despite worse cognition than controls, we tested the a posteriori hypothesis that WMH burden may be relatively more deleterious for ADHD than controls. We found considerably stronger negative correlations between total WMH volumes and Processing Speed (r=-.423, p=.009) and Executive Functioning (r=-.528, p<.001) in the ADHD group than in controls (r=-.231, p=.175 and r=-.162, p=.346, respectively), even though total whole-brain proportion of WMH (M=0.15%, SD=0.27; Mann-Whitney l/=430.0, p=.002) and frontal-lobe proportion of WMH volumes (M=0.33%, SD=0.51; Mann-Whitney U=464.0, p=.007) were lower in ADHD than in controls (M=0.29%, SD=0.42 and M=0.66%, SD=0.88, respectively).

WMH burden contributes significantly to the relationship between ADHD and cognition, but ADHD remains an independent contributor to worse processing speed and executive functioning in older adults. Vascular burden may have relatively more deleterious effects on cognition in ADHD, potentially due to decades of accumulated allostatic load, whereas healthy controls can accumulate greater amounts of WMH before cognition is impacted.