The main purpose in identifying modifiable risk factors for AD, such as IR, DM2 and cardiovascular disease, is to delay or even prevent neurodegeneration. Our results support this notion by postulating negative effects of IR on hippocampal volumes and cognitive performance in cognitively intact middle aged women at risk for AD. The importance of studying effects of IR on brain structure and cognitive performance is emphasized by the high prevalence of IR in the general population and its growth in direct association with rates of obesity.
IR might be a link between disorders of mood and cognition, as it may be leading to neuronal loss in the hippocampus due to the high concentration of insulin receptors in that region (Rasgon and Jarvik, 2004
). As summarized by Craft (2009)
, insulin transport across the blood brain barrier is reduced by prolonged peripheral hyperinsulinemia, which may help explain the observations of reduced insulin and brain insulin-signaling markers in the cerebrospinal fluid of patients with AD. In our sample of women at risk for AD, measures of IR were peripheral, consistent with other human studies using peripheral measures of IR as a proxy for central IR state (for example, (Craft et al., 1993
; Craft et al., 1998
It also appears that IR specifically affects the hippocampus, as we did not find an association between IR and total brain volume (r=.137, p=.342). Similarly, Convit et al. pointed to hippocampus as the main target for IR, which can explained by a preponderance of insulin receptors in that brain region (Convit et al., 2005
). In this sample IR was associated with verbal (ACT) and non-verbal (BVRT) measures with a brief one-trial stimulus presentation, but not with verbal learning/retrieval (BSRT) or nonverbal delayed memory (RCFT) where stimulus presentation occurred over a longer period of time. It is possible that IR may affect the very earliest and initial encoding process, namely acquisition of information, which may be an early indication of reduced learning capacity. Reduced learning capacity and defective encoding in patients with AD may contribute to the typical impairment of delayed memory (for example, please see Lowndes et al., 2008
; Pierce et al., 2008
). Therefore, it is plausible that observed differential performance in selective cognitive domains in relation to worsening glucose regulation/IR represent the first subtle changes in memory among middle-aged individuals at risk for dementia, which may further deteriorate with age and progression of metabolic dysfunction.
In line with our findings, Convit et al. reported a significant correlation of impaired glucose tolerance (as measured by the intranvenous glucose tolerance test) with both immediate and delayed paragraph recall, as well as hippocampal atrophy, in a sample of 30 non-diabetic, non-demented middle-aged and elderly men and women (ages 53–89, mean age of 69) (Convit, et al., 2003
). Our findings extend those of Convit et al. (2003)
, as they pertain to a larger sample (n=50) that is more homogenous with respect to age, gender, and other clinical characteristics, such as risk for dementia.
IR has also been associated with decreases in measures of global cognitive functioning (such as the MMSE) in several large population-based studies (Kuusisto et al., 1993
; Stolk et al., 1997
) (Kalmijn et al., 1995
; Vanhanen et al., 1998
). In this sample there was a decrease in both MMSE and WASI-FSIQ scores with increasing IR. In addition, both tests were highly correlated with each other (r=.374, p=.008). The WASI-FSIQ score is based on an abbreviated IQ measure that uses only two subtests, which are considered markers of crystallized intelligence; that is, they tend not to decline (at least not much) with decline in cognition, particularly the verbal subtest. Further, WASI-FSIQ is highly correlated with the full scale IQ obtained in the Wechsler Adult Intelligence Scale-III (The Psychological Corporation 1999
) and can be considered a measure of global cognitive functioning. Taken together, our findings suggest that decreased hippocampal volumes and measures of cognitive capacity and global functioning in association with IR may be harbingers for dementia risk, even while cognitive performance is within the normal range. While the association between hippocampal volume and cognitive ability in pathological aging may predict AD (Devanand et al., 2008
), in normal cognitive aging changes are likely to be more subtle (Bennett et al., 2006
), as in this sample of highly-educated, middle aged women at risk for AD.
Presence of genetic risk for AD by carriership of APOE-ε4 has been variably ascribed mediating role in the effects of IR on the brain (for example, please see (Craft et al., 2000
). In this study, presence of the APOE-ε4 allele did not alter the relationship between IR and hippocampal volume. Consistent with our data, data from den Heijer et al. on the association between IR and hippocampal and amygdalar atrophy in 929 nondemented individuals showed no differentiation with stratification by APOE-ε4 status (den Heijer et al., 2002
). Further, the degree of IR has been shown to be similar across APOE polymorphisms (Meigs et al., 2000
). On the contrary, the Honolulu Aging Study reported that the co-occurrence of DM2 and APOE-ε4 increased the severity of neuroanatomical abnormalities associated with AD, at least among Japanese-American men (Peila et al., 2002
). Among potential explanations for the discrepancy in findings among studies are different populations studied. For example, we studied only women HT users, whereas the Honolulu Aging study included only men.
This study has a number of limitations. As a part of a larger study on brain changes in the postmenopause, it was cross-sectional and as such, precludes inference of causality with respect to potential changes in hippocampal volumes and cognitive decline as a result of IR. Future studies are required to accurately establish the relationship between brain structure and cognitive performance, including cognitive decline, in relation to the metabolic functioning and should include both genders to identify potential gender differences in the relationship of IR with brain structure and cognitive performance. In addition, the effects of hormone therapy should be explored vis-a vis non-users to ascertain the potential interaction between reproductive markers and modifiable (IR) and non-modifiable (APOE-ε4) risk factors for AD.
Despite these limitations, these findings are thought provoking, as they suggest that in cognitively-intact middle aged women, IR is associated with decreased hippocampal volumes and diminished performance on measures of overall global functioning and in selective memory tests. Considering the importance of hippocampal integrity in memory, it is plausible that IR, as an independent metabolic state, may be associated with cognitive deficits and suggest a need for the early identification and treatment of IR in middle-aged persons.
In summary, these results are consistent with a growing body of evidence that links vascular risk factors to cognitive dysfunction in aging and provide additional evidence of the brain mechanisms that may mediate these associations.