Here we demonstrate that the AS task is a sensitive marker of executive dysfunction and frontal lobe structure in a well-characterized group of cognitively and functionally normal elders. Even after controlling for demographic variables such as age, gender, education, and ApoE4 genotype, the percentage of correct AS responses was related to executive function, most significantly, set-shifting on the modified trails task and generation on the design fluency task (, ). The percentage of correct AS responses plus self-corrected errors also predicted modified trails time. In contrast, the percentage of correct AS responses was not associated with performance on tests of language, visuospatial skill or memory. The percentage of correct AS responses correlated with gray matter volume in two nodes of the inhibitory control network: the right supplementary eye field (SEF) and left inferior frontal junction (IFJ). These results suggest that the AS task is sensitive to subtle frontal lobe dysfunction in normal elders that might indicate an increased risk of future cognitive decline.
The strong association of the percentage of correct AS responses (with or without self-corrected errors) with a variety of measures of executive function in normal elders is consistent with work in patients with neurological disease. We have previously demonstrated that patients with neurodegenerative dementia show similar correlations between AS task and neuropsychological performance, most strongly on executive function measures, 13
and similar correlations have also been identified in schizophrenic patients. 10, 15-18
As might be expected, given the greater variability in scores in demented subjects, the correlations between AS and neuropsychological performance are stronger and more widely distributed across cognitive domains in individuals with dementia than in normal elders.
Similar to our previous work in dementia, 13
we also identified a correlation between the percentage of correct AS responses and gray matter volume in both medial and lateral frontal lobe inhibitory control regions in normal elders. Both the right SEF and left IFJ gray matter volume were correlated with the percentage of correct AS responses; however, like the neuropsychology data, the strength of the association identified in the current analysis was smaller than that previously identified in dementia patients. The right SEF region that correlated with the percentage of correct AS responses is proximal to the right premotor region previously reported to be correlated with the percentage of correct AS responses in younger normal subjects (mean age 33 years). 48
Previous lesion studies have identified the DLPFC as critical for AS performance, but our analyses did not show a direct association of AS performance with DLPFC gray matter volume using a ROI based on published fMRI-derived coordinates. However, the IFJ region where gray matter volume correlated with the percentage of correct AS responses would have been encompassed by the lesions designated as DLPFC in a previous study that identified DLPFC as important for saccade inhibition in younger adults.22
Thus, the IFJ area we identified may have been involved in saccade inhibition in the lesion study. Together, the SEF and IFJ findings suggest that a low percentage of correct AS responses is associated with structural alterations within the same frontal lobe inhibitory control network nodes in normal older adults,
adults with focal cortical lesions,
and adults with neurodegenerative dementia, and that differences in severity of AS impairment reflect differences in the degree of inhibitory control network structural abnormality. We speculate that these structural abnormalities in our normal elders may arise either from genetic or developmental variation in frontal lobe structure or from degeneration of this network due to aging or incipient neurodegenerative disease.
Like a previous study that focused on normal elders, 31
we found no correlation between age and performance on the AS task with a 200 ms gap prior to stimulus onset. However, other studies have identified a relationship between age and AS performance. 29-32
In addition to differences in the AS stimuli that were used, several methodological considerations could account for the disparity between our findings and those from other groups. For example, previous studies did not account for education in their models and cognitive function was not rigorously assessed. Because age is a risk factor for Alzheimer’s disease, it is possible that some of the older subjects in these studies had subtle cognitive impairments that differed from the younger subjects.
Our findings support a possible role for cognitive reserve (CR) in maintaining executive function in normal elders. The CR hypothesis posits that cognitive decline resulting from brain damage can be limited by engaging other brain networks and mechanisms to compensate for a loss of function and that education is a surrogate for CR. 6, 49, 50
We found that the association of correct AS performance with executive function was most prominent in the lower education group (, ) Education did not correlate with the percentage of correct AS responses, suggesting that education alone cannot account for the observed variability in the AS performance in our subjects. The lack of association between AS and executive function in the higher education group therefore suggests that CR may play a role in mitigating the effects of inhibitory control network dysfunction to maintain executive function in normal elders. This might occur through greater compensatory activity or by recruiting of different networks in individuals with greater education and, by extension, CR.
Limitations of this analysis are the overall high levels of education in our subjects and the greater variance in performance in the lower education group for several tasks (modified trails time, modified trails errors and Stroop inhibition errors; ) that demonstrated a differential effect of education on the ability of the percentage of correct AS responses to predict executive function performance, suggesting that ceiling effects in executive function performance in the higher education group could explain part of this effect. However, this would not account for the effects of education on design fluency, Stroop inhibition or abstraction. More work using executive function tasks that are less vulnerable to ceiling effects in individuals with higher levels of education, and testing of a cohort with a broader range of education, will be necessary to confirm and better understand the potential role of CR in maintaining executive function in the setting of inhibitory control network damage.
We speculate that impaired performance on the AS task could serve as a marker of elevated risk of cognitive decline in normal elders. In a number of normal aging studies, declines in executive function precede age-related memory decline, 1, 3-5
however the mechanisms and clinical significance of these findings are not known. Since a lower percentage of AS correct responses indicates increased genetic risk of schizophrenia 51
and is a preclinical marker of Huntington disease, 27, 28
such findings in normal elders could indicate an alteration in frontal lobe structure due to a genetic risk factor or incipient neurodegenerative process, such as Alzheimer’s. By analogy to amyloid positron emission tomography scans 52, 53
or cerebrospinal fluid amyloid and tau levels 54
that can identify incipient AD pathology in normal elders, the AS task may be sensitive to frontal lobe processes that increase risk of cognitive decline.