Apolipoprotein-E Influences Aspects of Intellectual Ability in Type 1 Diabetes

A total of 96 people with type 1 diabetes were recruited and all completed the cross-sectional study protocol. Participants were selected from two preexisting cohorts (63 patients [9], 33 patients [10]), each of which had completed identical neuropsychological assessments. To minimize influences confounding neuropsychological performance, the following exclusion criteria were applied: hypertension (defined as blood pressure >140/90 mmHg), any previous central nervous system pathology, psychiatric disease, alcoholism or drug misuse, or multisystem disease known to affect the central nervous system.

The neuropsychological test battery was chosen to be sensitive to cognitive decrements across diverse cognitive abilities. Assessors blinded to the diabetes characteristics and APOE genotype of participants administered the neuropsychological tests in a fixed order. Incipient hypoglycemia, or hypoglycemia within the preceding 24 h, was excluded before neuropsychological assessment. Evidence of biochemical (blood glucose <3.5 mmol/l) or symptomatic hypoglycemia resulted in rescheduling of the neuropsychological session. The psychometric instruments used were as follows:

The Hospital Anxiety and Depression Scale (11) evaluated potential confounding effects of low mood and anxiety.

The Wechsler Adult Intelligence Scale-Revised (WAIS-R) (12) uses performance subtests to measure current intellectual performance (fluid and nonverbal intelligence) and are sensitive to disruption by organic brain disease. Four performance subtests were utilized (picture completion, object assembly, block design, and digit symbol tests).

The National Adult Reading Test (NART) (13) is relatively resistant to the effect of age and some types of organic brain disease. NART performance correlates more closely with premorbid IQ than demographic variables and was used to control for the confounding effects of prior intellectual ability (premorbid IQ, crystallized intelligence) in the present study.

Inspection time (14) was used to assess visual perceptual speed, a component of information processing ability. Participants discriminated between the spatial position (left or right) of the longer of two briefly presented vertical lines. The stimuli were backward-masked, the presentation duration was varied, and the duration of time required to reliably distinguish the stimulus (85% correct) was termed the “inspection time.”

Choice reaction time (15) was used to assess psychomotor speed and completed tests of information processing ability, and 1,2,4,8- and 8,4,2,1-choice reaction times were examined.

The Borkowski Verbal Fluency Test (controlled association) (16) is thought to assess frontal lobe and executive function. Participants have 60 s to state as many words as possible, beginning with letters of the alphabet specified by the assessor.

The Paced Auditory Serial Addition Task (17) was used to assess the ability to sustain attention and concentration. Participants listened to a number list, which they were required to add together according to a given rule. After practice, two consecutive 61-number trials were performed with 4 and 2 s between successive digits, respectively.

Severe hypoglycemia was defined as an episode requiring external assistance to facilitate recovery (18), and exposure was assessed retrospectively using a validated and formatted hypoglycemia questionnaire (9). Participants were asked to discuss their history of severe hypoglycemia with family, partners, or friends before completing questionnaires to improve accuracy of estimates. The total number of lifetime episodes, frequency of episodes, and total numbers of hypoglycemic seizures, coma, and episodes requiring medical intervention were recorded.

DNA was extracted from venous blood from all subjects using standard methods. APOE genotyping was performed using a PCR-restriction fragment-length polymorphism assay as described by Wenham et al. (19).

SPSS version 10.0 (SPSS, Chicago, IL) was used for statistical analysis. Demographic factors influencing neuropsychological performance (age, sex, and duration of diabetes), premorbid intellectual ability (NART), and the groupings of interest (severe hypoglycemia category, APOE genotype) were evaluated by multivariate general linear modeling (multiple analysis of covariance, MANCOVA). Severe hypoglycemia was dichotomized into two groups (group 1 = naïve to severe hypoglycemia; group 2 = previously exposed to severe hypoglycemia) based upon the total number of previous episodes. APOE genotype was categorized into two groups (ε4−, ε4+). MANCOVA was used to determine the significance (P) and magnitude (Eta²) of the variables’ effects on cognitive performance.

The clinical characteristics of participants are shown in Table 1. Data on glycemic control have not been included as the laboratory reference ranges, and the methodology for estimating glycated hemoglobin changed across the two time frames during which data from participants were collated.

Subdivision of participants by APOE ε4 produced two groups, ε4+ (n = 21) and ε4− (n = 75). The clinical characteristics are shown in Table 1. The subgroups had similar premorbid intellectual ability (NART) and similar exposure to severe hypoglycemia. Those possessing ε4 were slightly older (P = 0.033, t test) and tended to have had diabetes of longer duration (P = 0.059, t test).

The range of exposure to severe hypoglycemia was wide, from those with naïve to severe hypoglycemia (30%) to those who had experienced >10 episodes (29%) (Table 1). No significant difference in cognitive ability was observed between those previously exposed to severe hypoglycemia and those naïve to severe hypoglycemia (estimated marginal mean differences for those with and without a history of severe hypoglycemia are shown in Table 2).

The APOE ε4 allele was associated with a disadvantage in current intellectual performance (WAIS-R performance test score, P = 0.037, Eta² = 0.072) and a trend toward poorer frontal lobe and executive functions (Borkowski Verbal Fluency, P = 0.063, Eta² = 0.057) after consideration of age, sex, duration of diabetes, preceding severe hypoglycemia, and premorbid intellectual ability (Table 3). The relative disadvantage associated with APOE ε4 appeared to be sex-specific (APOE × sex interaction: F = 2.28, P = 0.04, Eta² = 0.15). Because of the sex specificity, MANCOVA thereafter was performed separately for men and women, using a similar model but excluding sex as a between-subjects factor. APOE genotype significantly influenced cognitive ability in women but not in men (Table 4). Women with APOE ε4 exhibited a significant cognitive disadvantage affecting current intellectual performance (WAIS-R performance test score, P = 0.001, Eta² = 0.26) and frontal lobe and executive function (Borkowski Verbal Fluency, P = 0.016, Eta² = 0.15). The difference in ability associated with possession of APOE ε4 in women was moderate-to-large (Eta²).

There was no statistical evidence of a detrimental interaction between APOE genotype, previous exposure to severe hypoglycemia, and cognitive ability (APOE × severe hypoglycemia interaction: F = 0.66, P = 0.68, Eta² = 0.05). The cognitive ability of subjects with the APOE ε4 allele, subdivided into those exposed previously to severe hypoglycemia (n = 15) and those with naïve to severe hypoglycemia (n = 6), is shown in Table 2. No significant difference in cognitive ability was demonstrated between the subgroups, although the numbers of subjects in this analysis are too small to be confident of a nonsignificant difference (Table 4).

The presence of the APOE ε4 allele was associated with a significant disadvantage in current intellectual performance (WAIS-R performance test score) and in frontal lobe and executive functions (Borkowski Verbal Fluency) in women with type 1 diabetes. The deficits did not encompass all aspects of cognitive ability, were present only in women, and were evident at a younger age (39 years) than that which has been reported in people who do not have diabetes (7). Female susceptibility to APOE ε4-associated cognitive disadvantage has been described for WAIS-R performance test ability in healthy elderly women (20), although the pathogenesis of the possible sex difference remains unclear. Age influences the effect of APOE ε4 on cognitive ability; neuropsychological performance in healthy children (21) and healthy young adults (22) is not affected by the ε4 allele, but middle-aged otherwise healthy adults (mean age 46 years) have been observed to have impaired learning and memory ability (7). The younger age (median age 39 years) at which ε4-associated cognitive disadvantage was observed in adults with type 1 diabetes in the present study implies premature susceptibility compared with nondiabetic individuals.

Laboratory and clinical evidence supports the concept of ε4-associated premature cognitive aging in diabetes. Elderly people with type 2 diabetes and the ε4 allele exhibit greater cognitive decrement (23) and an accelerated cognitive decline (24) compared with age-matched ε4 healthy control subjects; APOE from people with diabetes is irreversibly glycated and exhibits less in vitro bioactivity (25) when compared with nondiabetic control subjects. APOE bioactivity appears to have clinical relevance, as outcomes from a variety of cerebral pathologies are poorer in individuals possessing the least biologically active E4 isoform (8). As the present study did not include a healthy comparator group, the notion that APOE ε4 is associated with premature cognitive aging in type 1 diabetes is speculative at present, whereas this has been demonstrated in type 2 diabetes (23,24).

The present study did not support the hypothesis that polymorphism of the APOE gene may contribute to the susceptibility to hypoglycemia-induced cognitive decrement in people with type 1 diabetes. The number of subjects in each ε4 subgroup (n = 15 with previous severe hypoglycemia vs. n = 6 hypoglycemia naïve) does not give sufficient power to allow definitive conclusions, and further, appropriately powered investigation is required to evaluate this hypothesis. Irrespective of the influence of APOE alleles, cognitive performance was unaffected by severe hypoglycemia per se, consistent with the conclusions of the Diabetes Control and Complications Trial (3) and the Stockholm Diabetes Intervention Study (4), in which many participants had type 1 diabetes of relatively short duration, but conflicting with retrospective cross-sectional studies (2) that examined older subjects with diabetes of long duration.

In conclusion, the data in this modestly powered study suggest that APOE ε4 confers a cognitive disadvantage in young women who have type 1 diabetes. Further investigation is required to verify these findings and to determine whether the APOE ε4 allele is associated with premature or accelerated cognitive aging in people who have type 1 diabetes, the diabetes-specific factors that may be mediating any such disadvantage, and the possible interaction with sex.

S.C.F. was supported by funding from Eli Lilly & Company Ltd., and the study was supported by the Royal Infirmary of Edinburgh NHS Trust. J.C.E. received a Millhayes Scholarship from Exeter University. I.J.D. is the recipient of a Royal Society-Wolfson Research Merit Award.